Tracking Generators

[Eli Brandt]

Rich Holmes said:
> you can use it to convert, say, a linearly rising control signal to
> one that's flat for a while, then rises steeply.  But what would it
> do, if similarly configured, to a falling signal?  A flat one?  
> If I understand your interpretation, it's that it multiplies the
> control signal by another one which is continuous and piecewise linear.

I think (correct me if I'm wrong) that this is essentially what Ric's
design is doing.  The Oberheim tracking generator composes the two
signals rather than multiplying.  Think of it as a waveshaper for
control signals.  The tracker's five points define a response function,
which is applied to the input signal.

Let's say the tracker is configured into some zigzag.  A rising-saw
lfo will be reshaped into that zigzag repeated once per period.  A
falling saw will give the reversed zigzag, repeated.  A tri will give
first the forwards and then the backwards zigzag in each period.

lfo's are bipolar, -1 to 1.  You might expect that a unipolar envelope
(0 to 1) would only index into the right half of the zigzag, but
Oberheim realized that this wasn't too useful.  So the tracker maps
bipolar to bipolar and unipolar to unipolar (I think -- I'm a little
unclear on the whole unipolar/bipolar thing in the m6).  An analog
device won't know the range of the input signal, but a switch should
do the trick.

BTW, if you build one of these, think about whether you might want to
do uni to bi instead of uni to uni.  I'd prefer this myself, since it
gives more flexibility, and you can always just set the knobs into the
upper half of the range.

There was some discussion a while back about voltage-controlled
waveshapers and about a nifty homebrew module called a "scanning
generator" or something like that.  I don't remember if they had any
direct relevance...

> Your analog analog (if you catch my drift) sounds like a VCA driven by
> what might be called a slope sequencer: a sequencer programmed to put
> out a sequence of, not levels, but slopes.  Such a sequencer in and of
> itself seems like it'd be a nifty module.

Could you hook a CV sequencer into an integrator?  This would be more
modular, if it does what you want.

--
   Eli Brandt
   eli+ at cs.cmu.edu