Phaser designs: JFET, Vactrol, OTA, also parallel vs. serial?

[Digital Larry]

All phaser schematics I have seen cascade the phase shift stages.  This make perfect sense for a simple device where you want all the notches moving in the same direction and speed.  However I am conceiving something more elaborate where there would be multiple LFOs and envelope followers each controlling just one of the notches.  So one notch might be schwooping up and down at 0.2 Hz with a triangle, one might be wavering at 6 Hz square or saw LFO with very small width, one might be following an envelope going up and another might follow an envelope going down.   I think this might sound interesting.  I love phasing but am thinking is there a way to make it more subtle and complex at the same time?

Now by cascading the phase shift stages, you can't really make the notches go in different directions independently at the same time.  Any phase shift modulation introduced at a given stage would affect all subsequent stages.  But how about arranging them in parallel to a mixer stage instead of cascading them?  Then you'd tune the caps (most likely) to have each notch operate in the range of choice and yes you'd run the risk of notches overlapping but so what?  Then you could have individual CVs/LFOs for each stage and I think you could get some pretty interesting sounds that way.

The biggest problem I see is that with more than 2 phase shift stages mixing together in parallel at one point, you will probably eliminate the notch effect because you're not going to get 1:1 cancellations.  Just 2 stages tuned apart you could maintain it. 

===========
On a separate topic, is there any summary of advantages/drawbacks of using JFETs, OTAs, Vactrols?  I sense that OTAs might be the noisiest but easiest to deliver a repeatable result without resorting to part matching.  Also Vactrols are not exactly available any more (unless I missed something at Digi-Key or Jameco).

Thx for any insight.

DL

[Mark Hammer]

1) The need for vactrols will not go away for a long time.  Are they as availabl or as cheap as a 2N3904 or 10k resistor?  No, but they ARE fairly available.

2) Matching is critical for JFETs but not for other types of control elements.

3) Noise can be controlled.  It just depends how much you generate and what you are prepared to do to control it.

4) The intensity of the effect will depend on the number of stages in series, not the number of stages in total.  Placing a large number of stages in parallel, but few in series, will simply result in one or two notches occurring in unpredictable locations.  There is musical value to the unpredictability, but it is far better to have many stages in series with some sort of aperiodic modulation of the whole thing.  Two sets of 4 stages, each with their own LFO should be enough to produce that.

5) I built an envelope follower (albeit a not especially sensitive one) into my Boss RPH-10 such that picking control over sweep can be added to the existing internal modulation.  Go here and you'll find an example of what that can sound like:  http://hammer.ampage.org/

[R.G.]

All phaser schematics I have seen cascade the phase shift stages.  This make perfect sense for a simple device where you want all the notches moving in the same direction and speed.  However I am conceiving something more elaborate where there would be multiple LFOs and envelope followers each controlling just one of the notches.  So one notch might be schwooping up and down at 0.2 Hz with a triangle, one might be wavering at 6 Hz square or saw LFO with very small width, one might be following an envelope going up and another might follow an envelope going down.   I think this might sound interesting.  I love phasing but am thinking is there a way to make it more subtle and complex at the same time?

Yup. Been there. You're describing something of the underpinnings of building up a sitar bridge one notch at a time, as one example. The bridge on a sitar is shaped so the string curve as it vibrates interferes mechanically with the bridge to null/enhance certain harmonics. This is aside from the drones, but hey, mechanical vibrations are more complex than electronic. 

The reason they cascade phase stages is to get more notches. The limit is in a flanger where the notches are caused by time delay and you get a many notched comb filter with exponentially spaced (in frequency) notches. Phasers get one notch per phase shift of 180 degrees, so the cascading is to get as many notches as possible/practical, generally pooping out at two or three, although one notch phasers (MXR Phase45) and ten-notch or more phasers have been built.

Now by cascading the phase shift stages, you can't really make the notches go in different directions independently at the same time.  Any phase shift modulation introduced at a given stage would affect all subsequent stages.  But how about arranging them in parallel to a mixer stage instead of cascading them?

Parallel doesn't work, cascaded+independent should. To get a notch, you have to mix the phase delayed signal back with the dry signal. A phase delay by itself is an all-pass. And you can't run them in parallel - well, OK, you *can* run them in parallel, but the mixture of different-phase-delayed stuff from parallel stages mucks up the phasing and you get notches of odd and variable numbers moving in mathematically complex directions. What you can do is to make many one-notch phasers, mixing dry with phase delayed to get the transmission notch each time, then cascade those. A notch before a phaser removes signal content, and if it never gets into the next stage, it stays a notch no matter what the next stage does. So then you can drive each succeeding phase stage from an independent control signal and get more-or-less independent notch movement that is conserved from one stage to the next.

I think you could get some pretty interesting sounds that way.

Yep. Done this. It's just complicated for a guitar effect, so it doesn't get out much... 

On a separate topic, is there any summary of advantages/drawbacks of using JFETs, OTAs, Vactrols?  I sense that OTAs might be the noisiest but easiest to deliver a repeatable result without resorting to part matching.  Also Vactrols are not exactly available any more (unless I missed something at Digi-Key or Jameco).

What Mark said. I was about to refer to Mark and some of our previous discussions on phasers when I saw the note that he'd posted pop up.

The schemos are easy. The phase 45 is a good individual single-notch cell for this. By going series-independent, you can sidestep the washing-out effect of parallel.

[Digital Larry]

Thanks for the responses.  You know I am an EE by training but got into management 10 yrs ago and just recently got my mind going about doing some audio projects again and the old cobwebs are coming off, thus some of my dumb questions.

I also thought that you could use a crossover before 2 parallel stages mixing together with the dry signal, but you're right, just making the series stages independent with dry signal mixing before heading to the next stage accomplishes the same thing but MUCH easier.  Again it would be useful to experiment with tuning the notch frequencies between the stages to get an upper and lower range with independent CV sources.  I haven't ever cascaded 2 analog phasers, perhaps I'll grab a couple off eBay and give it a shot.  My experimentation jones has more to do with fancy CV/envelope generation (I LOVE the Adrenalinn) than making the phase shifters themselves.

Some of this is also inspired by the old Frank Zappa Micmix Dynaflanger legend, where he said "I set the envelope follower to track the high frequency decay which gives a more pillowy sound".  I'd also like to experiment with a flanger (not so much cascading flangers, what's the point? - maybe the crossover idea would be useful here) but to try the idea of taking a high-passed audio signal, getting the envelope, and using that for the phaser to get an animated but perhaps more subtle sound than pWOWp beyOWp bop-adoodly-twaing if you know what I mean.

[R.G.]

I am an EE by training but got into management 10 yrs ago

Yeah, been there. That can be either a good thing, for the higher pay or a soul-killer. Sometimes both at the same time. Periodically it becomes just a soul-killer, depending on conditions.

[Mark Hammer]

I've been offered management positions, and encouraged to go into it.  I tell people that it's not proper work for a grown-up with an education, and that I feel about managers the way Jane Goodall feels about chimpanzees.  She enjoys being around them, finds them endlessly fascinating and intriguing, and works tirelessly for their welfare and protection...but she has no desire to actually BE one.

Now back to the thread, already in progress.

[oldschoolanalog]

Now back to the thread, already in progress.

Phasers aside; 2 flangers in parallel make for some very interesting possibilities.

[StephenGiles]

And a phaser in the feedback of a flanger and vice versa also add for some interesting sounds, especially when they are controlled by opposite ramp CVs.

[StephenGiles]

And a phaser in the feedback of a flanger and vice versa also add for some interesting sounds, especially when they are controlled by opposite ramp CVs.

Also interesting when using a flanger is to have a ramp control voltage triggered only when playing harder, then taking priority over or joining the LFO CV.

[Mark Hammer]

And a phaser in the feedback of a flanger and vice versa also add for some interesting sounds, especially when they are controlled by opposite ramp CVs.

Also interesting when using a flanger is to have a ramp control voltage triggered only when playing harder, then taking priority over or joining the LFO CV.

I assume you mean a slow steady ramp such that the cycling of the sweep seems to move steadily upwards?

Although the insertion of fixed allpass/phase-shift stages into the delay path of a flanger has been productively pursued under the heading of "theta processing" as a way to space the lowest notches a bit further apart to yield more "musical sounding" flanging, I have not seen any mention anywhere of using swept allpass stages that might adjust that spacing contingent on where the lowest part of the sweep was.

[StephenGiles]

And a phaser in the feedback of a flanger and vice versa also add for some interesting sounds, especially when they are controlled by opposite ramp CVs.

Also interesting when using a flanger is to have a ramp control voltage triggered only when playing harder, then taking priority over or joining the LFO CV.

I assume you mean a slow steady ramp such that the cycling of the sweep seems to move steadily upwards?

Although the insertion of fixed allpass/phase-shift stages into the delay path of a flanger has been productively pursued under the heading of "theta processing" as a way to space the lowest notches a bit further apart to yield more "musical sounding" flanging, I have not seen any mention anywhere of using swept allpass stages that might adjust that spacing contingent on where the lowest part of the sweep was.

Ah ha, "compatability view works!! I messed about with swept all pass/phase shift stages in the delay path of a flanger back when I was drawing the Eventide Instant Flanger circuit. I could hear something interesting going on, but never pursued it. I think I may have been frog marched on to a plane for a trip to Argentina or somewhere else down there, and never continued the experiments afterwards!