Phasers: swirl vs. sway

[Benny]

Lately I've been thinking about what gives a certain type of phaser it's distincive sound, like the "sway" of a Univibe or the "swirl" of a Small Stone.  It would seem that there are two main factors: the shape of the LFO waveform, and the position and depth of the notches produced by the phase shift network itself.

I just completed a PIC-based LFO for an Easyvibe based phaser, so I can generate any arbitrary waveform that I can put in a lookup table.  I used equal caps (.01uF) in the phase shift network hoping to get a more swirly sound than the stock Easyvibe.  With a sine wave LFO, I was surprised how "Univibe-ey" it still sounded; there was definitely more high end swirl, but the distinctive lower end sway was still there.

The variable resistance elements I am using are Vactrol VTL5C3's.  As an experiment, I pulled the LDR side of the vactrols out of the circuit.  I turned down the LFO depth all the way, and adjusted the "drive" control (pot in series with the LED side of the Vactrols) until I measured 100k across one of the LDR's.  When I measured the others, I was surprised to find they measured anywhere from 80k to 160k!  Turns out these Vactrols are not very closely matched from the factory.  I had two extra VTL5C3's around, so I measured them all and picked the four most closely matched ones, which measured around 90k-110k at the same current.

I found that the phaser sounds much more swirly now, more like a Small Stone or MXR.  It's no secret that part matching is desired in JFET-based phasers, but it would seem that it helps in LED/LDR based phasers too.

So what did I learn from this?  Nothing that hasn't been discussed already I'm sure, but it seems that the more closely matched the elements in a phase shift network are (variable resistors, and caps and static resistors too I think) the more swirly it sounds.  If the components are not matched (out-of-tolerance LDR's, or intentionally mismatched caps as in the Univibe) the result is a more "swaying" sound.

Anyone come to the same conclusions here?

Benny

[puretube]

yes...

[RDV]

My experience is with the Phase 45(from Tonepad) which I recently built. I didn't electrically match the J-Fets, but rather plugged them in till I found a couple that would adjust around the middle of the 250k trim and sound good. I really dug the combo of sway & swirl I was getting. I then replaced the all-pass filter caps(both .047uF) with 10:1 ratio caps as per the 'Univibe Mod' and the sway & swirl I had going went away to a great extent, & I just had a high-filter/low-filter sound. My conclusion is that I had mis-matched the J-Fets(in a good way). That had already given me an uneven sound(in a good univibey way), so I replaced the 10:1 caps with the original values and my nice combo of swirl and sway came back. I can hardly turn this thing off, it sounds so very good, sort of between a Leslie cabinet & a Univibe.

Regards

RDV

[Mark Hammer]

Just about anything that degrades the focus of a filtering action will increase its "swirl" factor and make it more of an animation device than an obvious filter.  This is most easily noticed by adjusting the resonance on a phaser of flanger.  Cranking up resonance makes certan parts of the spectrum stand out more, and your focus drifts there.  Killing the regen and "defocussing" it gives it more swirl.  Mike Irwin came up with a very nice design for a phaser with more closely spaced notches than usual, so the focus is increased even without regen.  In stark contrast, most folks feel more swirl than filter from a flanger vs a phaser, precisely because attention is defocussed by the large number of notches distributed about the spectrum.

In the case of the Univibe, the swirl is increased even moreso by distribution of the notches a little more broadly, and making them a little more shallow.  The phasefilter mod to OTA-based phasers can also change if from a more filter-ey sound to more of an animated sound.

On my own Anderton Bi-Filter Follower, I have a range-shift switch for the upper filter, and even though it is still two staggered bandpass filters, moving them farther apart from each other still produces a certain defocussing (not to mention moving one filter out of the band where speech formants are) and increases the swirliness factor, relative to the typical envelope-controlled filter.

It's not just the depth,and distribution of any filtering that produces defocussing.  It is also the nature of the sweep.  Sweep your phaser with a blend of two desynchronized LFOs and I'll bet the increased aperiodicity will give it more swirl.  Who knows.  Maybe *mismatching* LDRs or FETs is important to getting swirl because the LFO will then not necessarily introduce the same amount of change in the control element at the same time while it sweeps (e.g., a 1v positive swing in the LFO signal introduced a 15% change in component state for LDR 1, a 25% change for LDR2 a 40% change for LDR 3, and a 50% change for LDR 4).  I'm not saying this is how it happens, but it could play a positive role and certainly doesn't have to play a negative one.

I *like* filter-ey, but I also like swirly, each for their own gifts.  I suspect this is more of a continuum than a dichotomy, and I think my examples demonstrate that.  Indeed, it would be a piece of cake to provide examples of effects with settings and sweep type here, in survey question form, and get everyone to rate the examples on a 7-point scale from "swirl-ey" to "filter-ey" and construct a hierarchy or plausible distribution.

[Joep]

Recently I build Tonepad's Ross Phaser. I made (after Francisco's suggestion) the capacitors switchable from stock, to the Univibe values, suggested bij Mark Hammer (see this thread).

It worked out really well, I can get al kind of sounds out of it, swooshy and swirly, the Regen control also helps (in the stock mode) to get some wild phasing going on. I also added the depth/intensity control after Mark suggestion.

All in all, this is a really versatile Phaser.

Bye,

Joep

[Chico]

I am not so sure that it is the shape of the LFO per se, that affects the swirly characteristic.  Although I have not done tests to confirm this, I think that it  is more the interaction of the LFO and driving element, and the position and depth of the notches produced by the phase shift network itself as you suggest.

Take the univibe type pedal.  I would imagine that, setting aside any "mojo cap values" etc. and component tolerance matching one defining characteristic of the swirly sound is the way the ldrs respond to the changes in light from the light bulb, which in turn, responds to the sine wave LFO.  Pull the ldrs and light bulb out, and replace them with optoisolators keeping the same LFO, and I would suspect that it would sound noticably different.  The way that optocouplers change resistance over time based upon a varying signal (LFO), I am guessing is different than the change in resistance of the optocoupler as the LFO varies, even if you were to match dark current and on current resistance values between the LDRs and optoisolators.  Moreover, the LFO on the vibe increases in amplitude as the frequency increases.  Still further, I believe that the sluggish response of a light bulb (compared to an LED in a common optocoupler) will have differing responses as the frequency of oscillation changes.  That is, at faster frequencies, the light bulb may not be able to fully shut off before the next LFO cycle.  As one final thought on this point, the light bulb is typically biased slightly on, even when the LFO depth is off.  

Compare the above with the manner inwhich the LFO works on your favorite phaser.

Therefore, my hunch is that one way you can make a piv lfo controlled phaser sound more univibish is to  model the behavior of the ldr/light/lfo interaction of your favorite sounding vibe, then you could use the pic to generate a custom wave form such that the interaction of the PIC LFO/optoisolator(fet) specific to your phaser design sort of  mimic the behavior of the LDR/bulb/lfo.  If you are trying to get away from the univibe sound, undue or ignore all of the above.

This is where I am heading in my research into this very question.
(Hey Benny, by the way, I am just finishing up my first pic LFO.  As a first attempt, I am using an ATMEL AVR processor and PWM output.  I have it set up so that I can vary the freq between 0.05 hz and about 30 hz and I currently have about 10 different "stock" wave forms programmed.   The design is still on the STK500 test board, but I hope to etch a pcb for it in the next week or so.  I would be interested in discussing your design with you.   I am sure that I am doing things way harder than need be)

[Benny]

Thanks for sharing your experiences everyone!  It's always good to know that you're not just imagining things.

Mark, IMO you are absolutely right that all these characteristics (swirl, sway, filter, etc) are not mutually exclusive.  The inherent design of the phaser certainly dominates the sound of the result, but it's interesting how the component selection and tolerance can play into this too.  It seems likely that mismatched variable resistors can have a "Univibing" effect on the sound; RDV, your experience with the the MXR Phase 45 suggests this, right?  In my case, I was saying to myself "why does my phaser still sound like a Univibe" and the culprit turned out to be the mismatched Vactrols.

I'm sure there is a lot more to this, I'm going to search the archives before I make people repeat themselves...

Benny

[Benny]

Chico,

You bring up a great point; the resistance across the LDR does not exactly follow the voltage out of the LFO.  In fact, the VTL5C3's turn off slowly enough to see with my DMM.  If you really wanted to digitally model a Univibe LFO, you would have to measure the LDR resistance at all speed/depth settings, and come up with some transfer function that you could apply to your basic sine wave generator.  Of course you would have to take into account the transfer function of the variable resistance element in your design as well.  In the end, you would have something far too complex for my lowly PIC to calculate on the fly.  Maybe it could be done with a set of lookup tables, which you select from based on the speed/depth settings.

Before the purists chime in here, I should remind myself: a Univibe is a Univibe.  A digitally modeled Univibe is a digitally modeled Univibe.   Not to be confused...

That's cool that you are working on a PWM-based LFO too!  I'd be interested to see what you have, and I'm happy to share my design too.  I'll draw up a schematic when I get some free time, and I'll see about posting it along with the PIC assembly code.

Benny

[Mark Hammer]

Benny, Chico,

It's bigger than you think.  I forget whether it was here or on Ampage, but there was a recent thread about the fuzz-phaser vs phaser-fuzz sequence and how they sound different.  Several of us noted that a phaser-fuzz order sounds more "Univibey" even though it is obviously NOT a Univibe.  It really is the damndest thing.  Try it out and I'm sure you'll agree.  It obviously doesn't sound *exactly* like a Univibe, but it really doesn't sound as pronounced as a phaser either.

So, why?  Following up on this idea I've proposed about the degree of "focus" in the filtering action, note that a phaser will *remove* frequency content as will any filter.  In contrast, a distortion will add frequency content.  Think of these as being like subtractive and additive synthesis.  Since the harmonic content provided by the fuzz is not limited by the phaser which precedes it (although it is *influenced* by the phaser), there is less "focus" to the sound.  The phaser shifts where all that extra harmonic content is distributed, but does not openly declare "Thou shalt only pay attention to THIS part of the spectrum" the way it would if it followed the fuzz, and filter out the high end or the mids as it sweeps around.  Moreover, since the shifting notches and peaks result in a highly variable degree of clipping across the spectrum (some parts clipping harder than others), you get this Jackson Pollock sort of spraying or scattering of harmonic content which is nonrandom but has less of a discernible pattern than if the order of the pedals were reversed.

Again, for me this illustrates that it is the diffuseness and reduced focus that makes the Univibe so pleasing at what it does, and whatever accomplishes the same kind of thing, albeit in another way, tends to feel very similar to us.

This is very fertile ground, and if somebody hasn't looked at it in any of the academic journals on music perception or psychoacoustics, it would be well worth looking into.