[sdiy] Configurable Phase Shift Project

[scottnoanh at peoplepc.com]

Hi Michael,

> With both phasers and flangers, there is always more apparent pitch
> bend at the low end of the modulation.  This becomes more apparent at
> faster modulation rates.  Adding lots of juicy regeneration/feedback
> seems to make this worse, unfortunately.  You can always minimize
> this unwanted pitch shifting by reducing the overall modulation
> depth, but where's the fun in that?  We all love deeeeeeep phasing
> and flanging modulation.

Yes indeed.  More power!!

> There are two ways that I can think of to keep the modulation fat and
> deep, but reduce the unwanted pitch bending:
>
> 1.  Use an LFO that changes its frequency (and hence, its shape) as
> it goes through its cycle.  Craig Anderton famously did this on his
> PAIA Hyperflange design.  He used a CEM3340 VCO for his LFO, and he
> fed a little bit of the LFO's triangle wave output back into the
> 3340's modulation CV input, so that the LFO would speed up during the
> triangles peak, then slow down at the low part of its cycle.  He had
> a special name for this type of LFO -- a "hyper-triangle", or
> something like that.  The end result was that you could use much
> greater modulation depths on the flanger, because the relative LFO
> speeds were better matched for the high and low frequency ranges of
> the flanger's comb-filtering.  Another way of describing this is that
> the modulation intentionally moves more slowly while it is working on
> the frequency range that provides the most dramatic "action" -- which
> is very useful with a deep, resonant flange effect, especially at
> very sloooooooow LFO rates.  I'm sure that this method would work
> just as well with a phaser.  One possible down side to this method is
> that the up and down modulation sound will have a kind of lop-sided
> quality to it -- this may or may not sound good at certain LFO rates.

I had seriously big problem with this long ago when I first started fooling
with phase shifters - I was using linear modulation.  Rene Schmitz suggested
I try expo control....much better.

This particular problem was two-fold: I was just forgetting that I was using
a very wide depth of modulation, and the rate had to be adjusted accordingly
(even a hypertriangular response will render the pitch shift at the low end
past a certain modulation rate).  The second part of the problem was that at
the *very* bottom of a wide sweep, even with fairly slow modulation, a
particular hump of shifted frequency would show up.  Decreasing the parallel
resistance with the LDR has smoothed out that hump with little sacrifice of
the bottom end response, and I'm pretty happy with it now (haven't posted
any samples of it yet).  This was what prompted me to go lower in
capacitance value before - what I was really doing was shifting the hump up
in frequency to where the ear didn't detect it with quite the alacrity it
would at the lower frequencies, where low pitch wobble is more easily
perceived as sounding 'un-natural'.

Between 10 mA and 1 mA LED current, the VTL5C3/2 moves from 2K through 11.5K
of resistance.  From 1 mA to .1 mA LED current, it moves from 11.5K to
around 900K, so it's certainly tempted to get up and boogie in the lower
ranges.  I think the expo control and making the lower end less dependent on
the actual LDR resistance value has overcome it.

I've used hypertriangular modulation before when working with flangers
(created the waveform by rectifiying a sine wave, flipping the upper half
over, then centering and mixing the two halves).  I was hoping to avoid that
with this design, as it would be nice to use modulation with more typical
waveforms, allowing easier external modulation.  Be that as it may, I'm
going to have to rig up a parabolic triangle and give it a go.  Especially
given the above resistance of the VTL5C3/2, it certainly makes sense.

As more stages are added, the more flanger-like it becomes.  Not really a
flanger of course, but more of a 'flaser'.

Thanks for the input!

Cheerio,
Scott