Multi-Notch Phaser - Theory Question

[polaris26]

Is it possible to make, say a three-notch (six stage) phaser, such that each notch frequency could be independently varied?  Would it be as simple as controlling each pair of variable resistances with its own independant LFO, or would you have to have three totally separate signal paths (essentially three independent two-stage phasers), and then mix the final outputs of these into one at the end (for example, three Phase 45 pedals with the same signal into all of them, and the outputs summed into an outboard mixer)?  I would think if you have only one phase-shifted ("wet") signal path, you could only vary the spacings of the notches, but the overall motion of the notches would have to track one another (such that you couldn't have, say the middle notch moving down in frequency while the low and the high notches are still moving up, etc) since the overall phase shift would still be the cumulative effect of all of the stages.  I hope this makes any sense?

thanks,
Dave

[R.G.]

Yes, it is as simple as a six stage phaser where each pair has independent control over the resistance of the phase for that pair.

There are two ways to do this, and I have not done the math to decide how one of them would act, that being the simplest one, just different LFOs to three pairs in a standard phaser circuit.

The simple way is with, conceptually, three phase 45s in series, not parallel. The first one makes a notch, then passes all of that signal along. The second impresses its notch on the signal it gets, and the third does as well. The notches move about in frequency as you'd expect them to.

The other way, putting independent LFOs to series pairs inside a single phaser with only one mix with the dry signal at the end is a bit complicated. What happens there is that each stage contributes more phase shift as the signal goes down the line, and only at the end when you mix the dry signal back do you get notches. You get notches at 180 degrees plus integer multiples of 360 degrees. Since all of the phases are additive, I don't know without doing some math whether it would have the same notch frequencies as the separate mixers case.

You could definitely get notches moving up and down at the same time this way, I just don't know whether it would be the same places that there would be notches in the case of three separate phasers.

[polaris26]

Yes I should have just put the Phase 45's in series - doh!  Anyway, I think this is the better way to do this, because I think when you have each pair of fets or LDRs controlled by a separate LFO, but all in one phase-shift network, when one lfo wave is ramping down and the other is ramping up, they are in a sense fighting each other for the overall phase shift at the end of the line (which is a cumulative effect of all the stages), so that yes you could have one notch going down and others up but I think the range of travel would be lessened due to this, and in no case could the notches overlap, whereas with the "three phase 45's in series" approach, you could actually tune each notch to overlap the other, or stay out of each other's way, depending on what you wanted. 

I was just trying to come up with something that sounds a bit thicker and less predictable than the standard phaser (after you hear the whooshing up and down, it sort of gets predictable) and I thought maybe having three separately controlled notches would be more swirly and chaotic sounding.

Dave

Yes, it is as simple as a six stage phaser where each pair has independent control over the resistance of the phase for that pair.

There are two ways to do this, and I have not done the math to decide how one of them would act, that being the simplest one, just different LFOs to three pairs in a standard phaser circuit.

The simple way is with, conceptually, three phase 45s in series, not parallel. The first one makes a notch, then passes all of that signal along. The second impresses its notch on the signal it gets, and the third does as well. The notches move about in frequency as you'd expect them to.

The other way, putting independent LFOs to series pairs inside a single phaser with only one mix with the dry signal at the end is a bit complicated. What happens there is that each stage contributes more phase shift as the signal goes down the line, and only at the end when you mix the dry signal back do you get notches. You get notches at 180 degrees plus integer multiples of 360 degrees. Since all of the phases are additive, I don't know without doing some math whether it would have the same notch frequencies as the separate mixers case.

You could definitely get notches moving up and down at the same time this way, I just don't know whether it would be the same places that there would be notches in the case of three separate phasers.

[gez]

The problem I encountered when doing this with allpass stages in series is that with independent LFOs, a lot of the phase shift that you create with one pair of stages can get cancelled out by the next.  End result, not a lot happening...

In the end I experimented with notch filters swept in quadrature.  Better results, but still not what I wanted (too subtle).  I should have figured out a way of trying to do it with BBD chips.  Clock Hetrodyning hell...

[R.G.]

Yeah, that's another reason to mix the dry back in after each notch-pair. That way, you fix the frequency notch in the signal. Then the next notch pair can impress its notch in the already-notched mixed signal. Once the mix with dry is done, a new notch-pair sees that as the dry signal and anything it does cannot cancel out what has already happened.

[polaris26]

Yeah - I think if I make something like this, I'm going to call it the "Buenos Notches" pedal.

Dave

Yeah, that's another reason to mix the dry back in after each notch-pair. That way, you fix the frequency notch in the signal. Then the next notch pair can impress its notch in the already-notched mixed signal. Once the mix with dry is done, a new notch-pair sees that as the dry signal and anything it does cannot cancel out what has already happened.

[gez]

Yeah, that's another reason to mix the dry back in after each notch-pair.

Yup, that was the solution I came up with (I'm sure I've mentioned all this in a past thread).  I still didn't like the end result, but I think I was expecting too much. 

What with all the mixers (I went 'active'), it seemed 'easier' in the end just to wire up (OTA) notch filters.  Lower parts count IIR, but don't quote me on that!