Multi-stage phasers

[StephenGiles]

Would 24 opamp stages or more be of any use in producing a very deep phasing effect or are we into Dome filters here - any views? - Mike, Mark, Ton, stm
Stephen

[Mark Hammer]

In some respects, given the number of notches/peaks produced by flangers at some points in their sweep cycle, you'd figure that phasers capable of generate many more notches/peaks than the typical 4-stagers would sound great and almost flanger-like.

The answer to that speculation is yes and no.  Mike has built, shown and demoed for me a 24-stager that sounds great, AND *almost* flanger-like.  A 16-stager too, IIRC.  Of course, when they are "demoed", we are talking about how the device cuts a swathe through white noise.  Apply the same device to a guitar through an amp, and one is likely to be less impressed.  Not to say that it will sound bad, but that the sheer majesty of that many audible peaks and notches tends to wash out and become more like what you'd hear from an 8-stager (which is still good, but obviously not the same).

So the question is "Why?".

For that, I turn to Marvin Jones' excellent 1978 Polyphony article regarding phaser/flanger differences, where he notes that the spacing  and number of notches in a flanger adjusts with time delay.  In other words, as the delay time increases (e.g., from .5msec to 12msec) in a flanger, the notches become spaced more closely, and also become more plentiful.  In the case of many-stage ("multi-" is probably a misnomer with phasers since one *does* always need at least 2 stages) phasers, the number of notches/peaks *available* does not change over the sweep cycle.  Why do I say "available"?  Because not all will be audible as the phaser sweeps in an "upward" direction.  Indeed, many will either be outsaide the range of hearing, outside the range of the speakers, or outside the instrument's natural bandwidth.  As the phaser starts to sweep downward, more and more of these *existing* notches will become audible.  Unfortunately, though, their spacing will not change over the sweep cycle as it does in a flanger.  They will be as far apart at one end of the sweep cycle as they are at the other.  This is one of the major differences in how each category of effect "feels* to the listener.

In some respects, there will be a *slight* overlap in feel since changes in the number of existing notches that become audible is *sort* of like changes in the number of notches produced (in the case of flangers).  However, unless one does something pretty fancy (2-pole allpass stages), the notches will rarely, if ever be spaced as closely as they are in a flanger at the longest-delay point in its sweep cycle.

There is also the matter of making sure that all 16 or 20 or 24 or more stages sweep equally.  Those who have built a Phase 90 can attest to how hard it can be sometimes to have FETs matched well enough to provide a nice sweep.  Certainly making a dozen or more FETs march to the same beat must be that much harder.  Which is why one will generally see OTAs used when there is a requirement to make 8 or more stages sweep (the MXR Phase 100 has 10 stages, but only 6 are actually swept), since they are easier to match.  It IS feasible to match FETs, and LDRs don't have to be that tightly matched, but 2-pole allpass stages can't make effective use of FETs the way 1-pole allpass can, and the idea of trying to illuminate 24 LDRs effectively starts to get pretty complicated.

IIRC, Mike's 24-stager was built using a quartet of 4049's as voltage-controlled resistive elements (surely you must remember the 6-stager project from ETI in the early 80's?).  These are easily matched but can only be configured as FET-to-ground for 1-pole alpass.  Mike worked hard to keep noise under control, and came up with some solutions, but I imagine I might have been considerably less impressed had the sample consisted of a gentle keyboard patch with lots of space between notes than the continuous white noise.  Great sound, mind you, but some things don't lend themselves well to performance requirements.

The bottom line is that, for most guitar-oriented applications, one is unlikely to achieve much audible benefit beyond 12 stages.  For synths, that ceiling can probably be raised a little, but you need to have lotsa bandwidth.

I wager that, once Mike chimes in, about 70% of what I say will turn out to be true and the remainder either poppy%^&* or just a tad off.

[StephenGiles]

Hi Mark, thanks for the input. You know I've occasionally heard the most unbelievably deep phased guitar sounds on a record and thought, well - you can't get that on a normal phaser pedal so maybe there's another explanation. Let's see what Mike has to say.
Stephen

[puretube]

without wanting to go too deep into details: feedback is an important factor...
:wink:

[Vsat]

Hi Guys,
Nice sunny weekend here (Mark - could you please contact me in advance if you are mentioning my name in a post - you do live only an hour away, no long distance charges... Thanks)

Just wrote a lengthy reply, phone line got disconnected and post got swallowed  
Will briefly mention major points:

Two main ways phasers are used:
a) for swooshes - helps to have dense wideband input signal spectrum for ear to pick out the moving comb filter response - if input signal doesn't have lot's of overtones, add some by distorting. A/DA Flanger and BF-1 and Jet Phase all sound boring with clean guitar when the distortion circuit (and  associated highpass) is disabled... there is a reason for the overdrive. The MF-104 Mooger Phaser is designed for easy overdrive. Similarly, tape flanging and TZF can be "difficult to hear" with clean, spectrally-sparse program material unless intentionally distorted.

Cont'd in next post in case I get cut off again...

[Vsat]

Cont'd:

The other way phasers are normally used is
b) to provide vibrato/chorus. In this case using "clean", spectrally-sparse input signal is fine. The ear does not pick out the comb filter envelope, but rather focuses on the resulting pitch modulation. Lots of stages helps here - the 24-stager can provide far more intense vibrato than the Univibe, or conversely, can provide a much slower chorus/vibrato than the Univibe for the same "depth". Or anything in between. The "allpass" vibrato/chorus can be richer sounding than a true time-delay BBD chorus since variable group delay is involved - the vibrato depth varies with frequency, effectively unlocking the overtone frequencies from the fundamental.

Cont'd...

[puretube]

http://www.beitec.com/phase/phse.html

[puretube]

http://www.beitec.com/digphas.htm

[Vsat]

Cont'd....

Another "trick" that can be used to make a sparse input signal into better "phaser-feed" is to apply a bit of vibrato before sending to phaser (NOT AFTER, as the benefits will be lost)... the few overtones present will be "wiggled" in frequency, and thus will "paint" themselves against the comb filter response for the ear to more easily discern...

For clean guitar, a  comb filter response with broad, hard-to-miss spectral features is probably easiest to hear... e.g. four to 6 stages...just a few widely-spaced peaks and notches, rather than a dense package of many peaks/notches. Add distortion, however, and the more peaks/notches you have, the more "Jet-like" it will sound... becomes more like a flanger, but with a different (less metallic) character. Top that off with smooth hi-regen behavior, and it is wonderful sounding.

But - I agree (along with at least one other phaser-builder on the sdiy list) that six-stages represents some kind of optimum size for the "classic sound", while  minimizing matching requirements, keeping construction simple, and maintaining high S/N without elaborate measures.
Regards, Mike

[Vsat]

Last one for now:

For an intense phasing sound without having lots of peaks and notches try broadening the notches... so they become more "trough-like" in form. For spectrally-sparse input signals like clean guitar or vocals this will provide a greater probability that one or more harmonics will be nulled out at any given time, and they will remained nulled out for a larger portion of the sweep.
Mike

[bwanasonic]

You know I've occasionally heard the most unbelievably deep phased guitar sounds on a record and thought, well - you can't get that on a normal phaser pedal so maybe there's another explanation.

Not sure what specific examples you have in mind, but adding compression after a phaser or flanger can make for some really chewy sounds. I started experimenting with these sounds after reading an interview with Frank Zappa where he discusses how he got the flange sound on "Ship Arriving Too Late To Save A Drowning Witch".

Kerry M

[Rob Strand]

To me, Phasers with a lot of stages start to sound like Flangers.

[puretube]

there are means, to prevent flangers from sounding like metal spring reverberators...

like Vsat & Bode say, that can be similarly used for phasing...

[lightningfingers]

I've built a 16 stage one...quite flanger-like, especially with feedback  Still looking for a suitable lfo though......

[StephenGiles]

Ton said:

without wanting to go too deep into details: feedback is an important factor...

Are we back to static phase shift stages in the feedback? Or shall I try my ADA Flanger wet signal in the feedback loop...............?
Stephen

[puretube]

the other way round: put your (at least 2 stages) phaser in ADA`s FB-loop
(wet only)...

[StephenGiles]

Ah, I'd better clear the decks then! Thanks Ton.
Stephen

[puretube]

or put`em befo` ADA`s wet, and mix ADA`s wet with the original dry signal...

[StephenGiles]

First option resulted in a great deal of squealing, much to the annoyance of my wife! So I'll try the second option.
Stephen

[puretube]

depending on the actual circuit, you`ll have to attenuate the backfed signal, or invert it...  :lol: