What do you do to produce "Theta processing"?

Started by Mark Hammer, July 22, 2009, 08:33:15 PM

Previous topic - Next topic

Mark Hammer

The now legendary Eventide Instant Flanger was distinguished by a number of things, like "bounce" in modulation, and envelope control.  Among the distinguishing features was also something referred to as "Theta processing".  You can read up a bit more about it at Jurgen Haible's site in the documentation for his Storm Tide flanger project: http://home.debitel.net/user/jhaible/jh_storm_tide_flanger.html

I have to say, though, that between his explanation, and the complexity of the schematic/circuit, I'm not getting it.  I can't tell if the additional 4 fixed allpass stages are added to the wet signal before mixing with dry, to an inverted signal used to produce sum and difference outputs, or what.  He emphasizes that it makes the biggest difference in stereo, but I'm not sure how it's implemented.

It sounds interesting, and potentially easy to incorporate as an add-on....if only I knew how Eventide implemented it.  Any with a little more clarity on the matter care to clear it up for me? 

Taylor

Well, it seems one channel is a BBD and 2 APFs mixed. The other channel is both BBDs and 4 APFs. I don't think there's anything special (with this block of the flanger) with inverting or anything.

I'm definitely interested in hearing this. Would be fun to build his circuit but I imagine the board would be monstrous.

Rob Strand

(Without adding my opinions:)  I only briefly skimmed over the article but the claim is that the linear spacing of the notches in a standard flanger causes the metalic tone.    A phaser on the other hand has more logarithmically spaced notches.   The larger idea effectively combines a phaser and a flanger (in series).  The simple form of this is to add fixed all pass filters in order to redistribute the notch positions.

I think the work "theta" implies some sort of phase modification (theta = angle = phase).

(I don't know if you have tried it, but a phaser with a large number of stages sounds like a flanger, with no metalic tone).



Send:     . .- .-. - .... / - --- / --. --- .-. -
According to the water analogy of electricity, transistor leakage is caused by holes.

vortex

Maybe it's a brain entrainment type thing which needs to be stereo to work and create the binaural beat frequency, in this case a theta state.   

Theta state is considered the deep relaxation range produced through meditation and other methods like brain entrainment by frequencies or rather the difference between two frequencies heard from the right and left side of the brain.

Theta waves are in the 3.5 to 7 Hz range, obviously lower than we can hear. By introducing a sound with a pitch of say 440Hz in the left ear and a pitch of 445Hz in the right ear you get a 5Hz wave inside your head. That's the binaural beat frequency. These frequencies can produce a state of deep relaxation/ meditation.

Just an idea regarding theta processing. Cool Edit Pro used to have a plug-in for this kind of thing!

Mark Hammer

But where do I insert the allpass stages?  I'm stumped.

slacker

From looking at the schematics it looks like it's arranged like this.
The dry signal feeds the BBDs, there's 2 of these in series and the signal is tapped off after the first stage for a short delay or after the second stage for a long delay. The signal into the second stage can be inverted. Feedback/resonance can be taken from either the first or second stage.

The dry signal also goes to the first allpass stage, which can be bypassed, it then gets mixed with the short delay from the BBDs, and goes to one output
The output from the first allpass stage also goes to the second allpass stage, which can also be bypassed. The output of this gets mixed with the long delay from the BBDs, and goes to the other output.
I don't quite get how the left and right switches are supposed to work, but it looks like you can assign either output to the left or right channels. I'm not sure what the mono signal consists of, I assumed both the outputs would be mixed together but I can't see where or how that happens.

Don't know if that helps at all :)

Mark Hammer

I just popped off a note to Jurgen himself.  Perhaps he'll be able to clear it up.

What intrigues me and piques my interest is the possibility of a small daughterboard that could be used to adapt any of a variety of DIY and commercial flangers.  First, I have to know how it works, though.

Rectangular

interesting read. I'll keep my eye on this one.  I feel like I've built just about everything... running out of new, interesting circuits to hold my attention :D

Mark Hammer

Update.  Starting to make sense now.  At least one of the lights has gone on :icon_biggrin: , courtesy of one of the forum members (whom I would credit here, but I can't access the e-mail account who contacted me on from work right now), and the inimitable Bernie Hutchins.

One of the critical differences between flangers and phasers is that, where the number of notches produced by a phaser remains fixed according to the number of stages, as a flanger sweeps "lower" (i.e., adds more delay) the number of notches increase, with the notches being very closely spaced in the lower part of the spectrum, and a tank-y or box-y or metallic-sounding toen resulting if one introduces any resonance whatsoever.   To be fair, in more complex phaser designs like 16 and 24-stage phasers, many of the notches produced are largely inaudible until the unit starts to sweep lower such it seems like there are more notches as it goes lower.  But for your meat-and-potatoes 4-6-stage phaser and your meat-and-potatoes 1-12msec MN3007/3207-based flanger, that's how it works.

The "theta" idea attempts to essentially stretch out the spacing of the flanging notches in the lower realm by applying a frequency-based phase-shift to counteract the notch-spacing produced by time-based phase shift. 

What made the lights go on for me was when I finally  :icon_rolleyes: noticed that the phase-shift stages shown in the Bernie Hutchins paper and the Storm-Tide schematic had the cap and resistor reversed. :icon_biggrin: :icon_idea:  Normally, in the MXR Phase-90 type pĥase-shift stage we are all accustomed to seeing, there is a cap between the previous stage and the non-inverting op-amp input, and a variable resistance (LDR, FET, or other) to ground.  That little network forms a high-pass filter.  As the resistance is varied, the 90 degrees of phase-shift attainable by that stage is applied either higher-up or over more of the spectrum.  However, at all times, the least phase shift is applied to the lowest parts of the spectrum.  If you flip those two components around, such that the cap now goes to ground, they form a lowpass filter, and the potential phase-shift from that stage is always maximal at the lowest frequencies and minimal at the highest ones.  I always forget which is called which, but the terms used are "phase-lead" and "phase-lag".

Theta processing involves adding some fixed phase-shift to the lower portions of the delayed signal prior to mixing with the dry signal.  The net result of applying both frequency-based phase-delay (from the allpass stages) and time-based phase-delay (produced by the BBD) is that the notches are spaced farther apart at the low end when the flanger sweeps downward.  In principal (and apparently in practice), this will sweeten the flange sound and reduce the irritating metallic quality at lowest sweep points.  I am VERY slowly wading my way through the Hutchins and Haible material, but I think this is the gist of it.

What it suggests is that there may be "reduced" versions of theta processing that could be applied to DIY and commercial flangers, in the form of a small add-on board.  That add-on board could conceivably be between the delay path and mixing stage, or it could conceivably be in the recirculation path only such that it really only kicks in as the regen is turned up.  A "reduced" version might be as trivial as two stages of unity-gain fixed allpass between the regen knob and the return point on the main board.  Use a SIP chip and vertically-mounted resistors for that, and it could be conveniently inserted ibn plenty of commercial pedals....assuming it does the trick.

At this point, I'll shut up and let the more mathematically inclined pitch in.  But I think I just found something to incorporate into my ever-evolving PAiA Hyperflange!