Interesting effect with two phasers in series

[Mark Hammer]

One of the delights of DIY is that you can inexpensively make multiples of things.  One of the delights of true bypass switching (compared to FET-based) is that modulation pedals that use dry+wet for effect, can have dry cancelled and still permit proper bypass of the whole circuit.

Last night I took two 4-stage Ross phasers and placed them in series, set to dissimilar slow wide sweeps with lots of resonance.  Very nice and spacey.  Steve Hillage would have been proud.    Then I switched the first one from phaser (dry+wet) mode to vibrato (wet only) mode, and the strangest thing happened.  It turned into a very slow tremolo.  The second unit still had audible sweep, but the volume of the overall effect faded in and out in conjunction with the LFO rate/width settings, by losing bandwidth from the bottom up.

In retrospect, it made sense.  Here, I had (if you ignore the separate chassis and think of it like one circuit) 360 degrees of phase shift, followed by another 360 degrees of phase shift.  After 360 degrees, the signal was essentially split, meeting up again later on.  At 720 degrees, the 720-degree version of the signal was recombined/reunited with the 360 degree version.  But keep in mind that most phasers operate by applying phase shift in a manner that moves the point where maximal phase shift (90 degrees per stage) occurs up and down the frequency spectrum.  As the first phaser sweeps downward. more and more of the complete spectrum is 360 degrees out of phase with what it re-unites with after the 8th stage in the second phaser.  Neat, huh?

I did this with separate LFOs in separate pedals.  What I'm curious about is what would happen if the two units were synced to the same LFO.  I have a board with a Univibed Ropez and Normal (well, as normal as *I* can make it) Ross phaser, each with separate ins and outs, but intended to be installed into a box where I could slave the two to a common LFO and cascade all 8 stages.  Like any good Univibe, mine has a chorus/vibrato switch, so if I set it to vibrato mode and feed its output to the phaser with the two units fed by a single common LFO, I should be able to find out what happens.

Incidentally, this suggests that if you feed each channel of a stereo amp with a phase-shifted signal (wet only), you should get not only notch-type cancellations but also interesting vibrato and volume modulation effects.  Take your Gravol kids!

[oldrocker]

That's sounds very interesting.  I always wanted to plug two phasers together to hear what would happen but never had two at the same time to try it.  I knew something good would come out of that.  I haven't built a 4 stage phaser yet but I plan on it and maybe I'll build two.  I did build a Phase 45 with Univibe mods and just got it working right today as a matter fact.  I love that pedal now.  I was testing it out and my keyboardist stopped by and I was using the P-45 with my tremolo pedal (Pulsar).  He got really excited when heard the two pedals together.  With two different speed sets between the two pedals he said it was the Leslie effect he's been looking for.  He asked me if there was a way to put both effects in one box with foot operated speed controls and he would pay me.  I don't know if I want to get into that since so far I perf board everything and that's a lot of work.  Maybe some day I'll try something like that. 
For the two Phaser builds I'll purchase a couple of PCB's and go that way.  Too many parts for a perf build.  Anyway two phasers hooked together hmm... that would be fun to play with.

[bioroids]

It does sound very interesting!

In retrospect, it made sense.  Here, I had (if you ignore the separate chassis and think of it like one circuit) 360 degrees of phase shift, followed by another 360 degrees of phase shift.  After 360 degrees, the signal was essentially split, meeting up again later on.  At 720 degrees, the 720-degree version of the signal was recombined/reunited with the 360 degree version.  But keep in mind that most phasers operate by applying phase shift in a manner that moves the point where maximal phase shift (90 degrees per stage) occurs up and down the frequency spectrum.  As the first phaser sweeps downward. more and more of the complete spectrum is 360 degrees out of phase with what it re-unites with after the 8th stage in the second phaser.  Neat, huh?

I dont quite understand why is it happening anyway, the 360 degrees out of phase shouldn't have a noticeable effect, as you need 180 degrees for cancelation.

Maybe the high resonance has some influence in this. Have you tried it with the resonance at zero?
In any case, I think this can lead to interesting new effects!

Thanks for sharing it 

Miguel

[Mark Hammer]

I guess I confused things.  Let's describe it this way: to produce cancellation, the phase shifted signal always has to be 180 degrees out of phase compared to something.  Normally, that "something" stands still, and the cancellations only occur where the cumulative amout of phase shift across the 4 (or more) stages adds up to 180 degrees, or multiples.  Normally, the amount of phase shift introduced by each stage is pretty minimal WAYYYYYY down deep in the spectrum, so that even when you add it all up across 4 stages, you never quite get enough phase shift to result in cancellation against the dry signal.

It appears, however, that somehow the "pre-phase-shifting" of the dry signal ends up producing something that yields full-spectrum cancellation.

This is one of the many reasons why I am eager to try the experiment described with the Univibe (in vibrato mode) and phaser in series.  The Univibe, because of the staggered cap values, produces a much flatter and even distribution of phase shift (well, not quite as focussed, actually) across the spectrum.

Other weird thigs you can try out include this.  In traditional op-amp/FET-based phasers, hase shift is applied from some frequency upwards when the cap straddles the stages and the FET goes to ground.  When the variable resistance straddles the stages and the cap goes to ground, though, the phase shift is applied from some frequency downward.  Okay, so take your dry signal, take one or two "flipped" phase shift stages and insert them between the point where the signal gets split into wet and dry and the mixing stage so that there are two fixed phase-shift stages of "dry" signal that get mixed with 4 stages of variable/swept phase shift.  I wonder what would happen.

[R.G.]

Think of it this way. One phase shift stage (not a pair) has a 90 degree phase shift at some point determined by the phase R and phase C. The actual phase of the signal out of the phase stage starts departing from the original signal at a frequency 1/10 of the 90 degree frequency, and closes in on 180 degrees asymptotically at a frequency 10x the 90 degree frequency. When you cascade a number of these stages, you get a setup where the output signal has a phase shift compared to the unshifted input that increases (or decreases, same thing) with frequency at rates and amounts depending on how many stages and the tuning of each stage. The phase delay is not a fixed amount at any one frequency, but changes with every frequency.

If you stagger these carefully, you can get a linearly increasing (or decreasing, same thing) phase difference from the input signal. Two of these stagger tuned will exhibit an output that is, for instance, 90 degrees apart at every frequency compared to each other. This is the basis of a dome filter or pseudo Hilbert transform filter.

What is happening with two phasers is that they both have a certain profile of phase change per frequency; for instance, the phase may increase by 90 degrees ever 10x frequency over most of the audio range. Compared to a static signal with a constant 0 degree phase shift, they come into and out of phase. When mixed, the dry and either wet signal will reinforce and cancel at different frequencies as the phase changes.

But two phasers, with phase shift increasing in nearly the same way with frequency will have a very similar phase versus frequency. By changing an LFO to walk them into reinforcement, quadrature, then cancellation, all frequencies cancel at the same time, because they're all shifted relative to 0 by fixed amounts. Two identical phasers would be needed for a true tremolo.

The frequency-dependent nature of the tremolo is a characteristic of the two phasers not being >> exactly  << identical, so some parts of the shifted outputs come into reinforcement and cancellation at different times.

[StephenGiles]

Wet only from 2 phasers into a stereo amp - Mark I tried that many years ago and through headphones it sounded unbelievable!

[Mark Hammer]

Here's a stereo trick to consider trying. 

Imagine one had 2 or 4 stages of the R/C type allpass stage (i.e., variable R is between output of previous stage and + input of op-amp, with cap to ground), and the same number of C/R stages (variable R goes to ground, and cap links op-amps).  Keep in mind that the one increases the amount of phase shift per stage (up to 90 degrees) downward while the other incrases the amount of phase-shift per stage upwards.  As the various resistances got smaller (assuming they are all of the same type, such as LDR, and controlled by the same LFO), the point of maximal phase shift for each stage (where it "plateaus" to 90 degrees beyond that frequency) would change synchronously.  So, if I've understood things correctly, an R value of 10k and a C value of .01uf in all stages would produce 90 degrees of shift per stage below 1590hz in the one type of allpass, and 90 degrees of shift per stage above 1590hz in the other.

Correct so far?

If you tried to mix the two, as if they were a dry+wet combination, my guess is you wouldn't get anything terribly dramatic since they more or less track each other.  But what if each allpass path were mixed separately with a dry signal and THOSE signals formed your stereo outputs?

[StephenGiles]

Again Mark, I've tried that and it sounds amazing. I think it was with 2 Badstones - one original and one I built. The same goes for flangers connected in the same way, and using a Rat before each........even better!

[Mark Hammer]

Well, my wife will be out of town next week, the football season is getting uninteresting, and our older will be moving out soon, so I may finally have some time to cover all my obligations AND try some building experiments.  You've certainly tempted me.