[sdiy] Glide question

cheater cheater cheater00 at gmail.com
Mon Sep 6 13:47:56 CEST 2010

Hi guys!

Thanks for the amazing replies. This thread is very informative.

On Sat, Sep 4, 2010 at 21:14, Antti Huovilainen
<antti.huovilainen at iki.fi> wrote:
> The other alternative is linear glide, where the CV moves at a
> constant rate (possibly recalculated for each new note) towards the
> target value. You could achieve this by using one-pole OTA lowpass
> filter and heavily overdriving the input, so that the tanh() clipper
> limits the rate of change to Ictrl.

Thanks for the reply, Antti. I wonder which glide would be better for
a step sequencer. I will try to find a synth with linear guide and see
how it works.

On Sat, Sep 4, 2010 at 21:36, Olivier Gillet <ol.gillet at gmail.com> wrote:
> There are two ways of doing glide/portamento:
> - "constant time" in which sliding between C3 and C5 takes as much
> time as sliding between C3 and C4.
> - "constant rate" in which sliding between C3 and C5 takes twice as
> much time as sliding between C3 and C4.

I think what I want is "constant time"...

On Sat, Sep 4, 2010 at 22:33, Tom Wiltshire <tom at electricdruid.net> wrote:
> an RC filter (VC or not) gives you constant time. An integrator (again, VC
> or not) would give you constant rate. There is one other difference between
> them - one is a cap charging curve, and therefore the glide slows down as
> it approaches the target note, whereas the other is linear wrt to pitch and
> glides at the same rate all the way to the target.

Thanks. I'm actually not sure which one I would want for this. The RC
filter gives you the cap charging curve, while the integrator gives
you the linear response.

I guess you can get a constant-time linear glide too, since you know
the endpoints - the values of the steps you are gliding through - so
that should be easy (for someone who knows what they're doing, i.e.
not me!)

The idea with the asymmetric RC filter is also very good. I should
check it out! How would that work with an integrator (for asymmetric
linear glide)? Is there such a thing as an asymmetric integrator?

What shape do you guys think is better for glide on step sequencers?


On Sun, Sep 5, 2010 at 09:15, David G. Dixon <dixon at interchange.ubc.ca> wrote:
>> > A few examples of glide circuits, based on Harry Bissell's "Morphlag"
>> > which allows independant up and down rates, and can vary the glide
>> > shape from linear to log.
>> > http://www.wiseguysynth.com/larry/schematics/Morph-Lag.htm
> I've just been playing with a simulation of this circuit, and I have a few
> questions (Harry?):
> 1.  Is there a good reason why the shape-controlling feedback resistors are
> twice the input resistors (i.e., 20k instead of 10k)?  If they were equal,
> then the whole circuit just be a first-order lag with TF = 1/(1 + sRC) at a
> 100% (i.e., fully exponential) shape pot setting.
> 2.  Is there a good reason why one doesn't simply return the opamp output
> directly to the shape pot through a single resistor, rather than the two
> diode outputs through two resistors?  My simulations suggest that this would
> work equally well, if not better.
> 3.  According to my simulations, a small resistor (say, 22 or 33 ohms)
> between the shape pot and ground helps to damp oscillations at the opamp
> output which would otherwise occur at a zero shape pot setting (i.e., fully
> linear).  These oscillations are worse if the diode outputs are fed back
> rather than the opamp output directly (see comment 2).

More information about the Synth-diy mailing list