Phaser question....

[digi2t]

I've noticed that phasers have "stages". What do more stages provide sound-wise. 4 stages, vs. 8 stages for instance? Is it "swooshier" (hope I didn't infringe on anything Nike related  ). How many stages would be uber-cool?

Never was a big phaser user per se, but since my FK-2 project, it's got me wondering.

Thanks.

[R.G.]

http://geofex.com/Article_Folders/phasers/phase.html

[digi2t]

http://geofex.com/Article_Folders/phasers/phase.html

R.G.... you nasty, nasty man. Unveiling the answer to my question, at the very end of your article  .

I thank you, once again.

[Mark Hammer]

More stages results in more notches.  So, 4 stages = 2 notches, 8 stages = 4.

More notches-in-theory does not necessarily equal more audible notches or a more "intense" effect.  There has to be spectral content where the upper notches are in order for there to be an audible dip in that content.  So, if a person had a 24-stage phaser, it would be largely indistinguishable from an 8 stage if you were feeding a clean bass through it.  In that respect, what applies to flangers also applies to phasers: they sound "best" when feeding them with multi-source broad bandwidth content.  Mike Irwin demoed a 24-stager of his for me once, and it sounded fabulous with white noise as a signal source.  Had I plugged my guitar nto it, there would have been precious little content above maybe 4-6khz and the upper 6 notches produced would have likely been beyond the range of guitar signal.  That is, of course, why the 12-stagers appeal to synth players who DO use white noise and other harmonically rich signals where the HF content lingers long enough for a sweep to be heard.

For guitars played through guitar amps, there is diminishing return above 8 stages.  You can probably hear a slight difference between 8 and 12 stages, but it would depend very much on tone settings.  The difference between 4 and 8 is clearly audible, independent of tone settings, though.  I have a Boss RPH-10 that does 12 stages, and one if hard pressed to hear what the extra stages add.  It's a bit like the change in performance between a netbook running Win XP or 7 with 1 Gig of RAM, vs 2 Gig, vs 4 Gig.  The jump in performance from 1 to 2 Gig will likely be much more noticeable than the jump from 2 to 4.

In the case of flangers, the most notches are produced when the sweep is at its lowest point, such that even if we maybe don't get to hear ALL of the notches produced, we still get to hear a lot of them, spaced close together.

Mike also demoed me a phaser design of his whose engine I cannot reveal, but it produced notches that were more closely spaced and deeper than we are accustomed to.  I seem to recall that version only produced 3 notches, but it was an extremely "vocal" sounding circuit.  I am confident that if a larger version was made to produce 4 or 5 notches, it would have produced some very robust sounds.

Though I've only done the experiment with two additional fixed stages in one case, and four additional swept stages in another - both instances of OTA-based circuits - I find that those with more stages don't seem to sweep as high.  Don't know what it is but when I built a Ross/Ropez and nudged the current-limiting resistor from the LFO to OTA stages from 10k down to about 8k2, I was getting this amazing sweep that went much higher than I had ever heard in a phaser.  Not as "swooshy", but certainly as dramatic a sweep from a phaser as one might hear in an A/DA Flanger (i.e., wider than wide).  I wired in a little daughter board with two op-amp based fixed phase shift stages, and while the sound felt thicker, it didn't seem to sweep nearly as high.  Same thing when I wired up a 4-stage expander board for the Ropez.  Don't know what it is.  In the latter case, it may be the sharing of the same current by more OTAs, but I think I reduced the current-limiting resistance to compensate.    More resonant, certainly, but just seemed to do "the turnaround" at a lower point and I don't know why.

So a question that I never pondered before: when one adds another pair of stages to produce an additional notch, where is that notch added - above the existing ones, or below?

[R.G.]

Hmmm.  I wonder what happens when you take two-stage/one-notch phaser sections, do the summations after each stage, then drive the notches separately, either in sync, inverse, or randomly. 

[Mark Hammer]

Should one assume the two section in each cluster have the same specs, such that, fed with the same control signal they'd produce the same notch at the same point in the spectrum?

[digi2t]

So, in a nut shell, if I understand this correctly, for all intents and purposes having a mole of stages is absolutely pointless. In real world guitar application, the human ear can only distinguish sounds within a certain frequency range, so exceeding that range isn't really useful right? OK, I can handle that. When you mentioned synths, it made complete sense.

Mike also demoed me a phaser design of his whose engine I cannot reveal, but it produced notches that were more closely spaced and deeper than we are accustomed to.  I seem to recall that version only produced 3 notches, but it was an extremely "vocal" sounding circuit.

Oh PLEASE PLEASE PLEASE... just a little peak! When Jimi and I hear "vocal", we tend to drool! 

Thank you gentlemen for the wonderful lesson. Have yourself a great day. 

[Mark Hammer]

Oh PLEASE PLEASE PLEASE... just a little peak! When Jimi and I hear "vocal", we tend to drool! 

Nope.  Solemn vow in place.

As for the audibility, it's not so much restricted by human hearing as it is by the absence of anything to actually hear.  If the signal to be processed was simply drenched in lots of content between 8khz and 18khz, and the speakers themselves could reproduce it (AND the listener hadn't already destroyed their cochlea with ear buds), then having notches imposed in that region would make for an audible effect.  But if the guitar signal doesn't really go up there, and the speakers don't accommodate it, then whatever is "going on up there" is within human hearing range, but not audible in terms of its impact.  A bit like listening to the impact of putting new strings on a guitar when the treble is turned down full and the amp is covered in pillows; something may well be "going on", but damned if I can tellN

And for the synth thing, remember that the brunt of the harmonic content generated by a guitar string is within the first few 1/10ths of a second, and after that we're quickly back down to fundamental and a few lower-order harmonics.  So even if one uses SC pickups and the amp is flat out to 20khs, and feeds a piezo-tweeter-equipped amp, whatever there is to be notched "way up there" disappears quickly, unless you're picking your brains out.   Fuzzes can help to provide more to filter, but even they have their limits.

Finally, it bears noting that a good part of what a phase shifter does is redirect your attention.  It does so by having a focussed impact on the treatment of the signal.  That's why a mere one or two notches can produce such a striking effect: because it's targetted and not spread out.  A flanger, on the other hand, when not used as a sort of chorus, creates part of its impact by "infecting" the signal with an increasing number of notches.  Some of those notches ARE more obvious than others (which is where the "metallic" quality of them comes from), but where flangers sweep from barely present to over-running the signal, phasers sweep from "up there" to "down here".  It's a different sort of psychoacoustic or perceptual effect.

And of course, vibe units do what they do by removing that focussed effect.  The multiple cap values in what is otherwise a normal 4-stage phaser, produces a very broad and shallow dip, rather than a more focussed notch (again, I thank Mike Irwin for some graphical data that provided insight).  because those dips are fairly broad and shallow, our attention is not directed to a targetted impact that moves up and down in the spectrum.  Instead, we just perceive a sort of undulating animation; at least when it is done well.  never seen a 6-stage vibe, though.

Having said that, if one had a 12-stager, and 3 or 4 of the notches produced were up in the stratosphere (i.e., in audible) during the upper swing of the sweep, then for all practical purposes the phaser would sound like it is going from fewer to more notches as it sweeps downward.  However, as RG's document indicates, the notches don't change in their spacing, as they would with a flanger.  Again, even though one IS, in effect, going from fewer to more notches, the spacing issue makes for a different psychological consequence.

[pinkjimiphoton]

I'll sign away a disclaimer...pretty please?

don't wanna hear your bud's thing, but would love to hear more about yours!

that didn't sound right...lol.

dino's right tho...we're formant addicts!!!

[R.G.]

Should one assume the two section in each cluster have the same specs, such that, fed with the same control signal they'd produce the same notch at the same point in the spectrum?

Depends. As you note, the audibility of a notch depends a lot on both where it is in the audio spectrum and on the content of the signal it's fed. And it also depends on how you want to drive the sections - what control voltages/currents you have available or are willing to generate.

If you want more than one notch to always move in synchrony but at different frequencies, it makes sense to feed them exactly the same control signal, and install a permanent offset with the timing caps.  If you're going to do moves-opposite-directions, you have to do that with the control signals and may as well make the timing caps and other time constant stuff identical. Likewise, with random notch movement, there's no advantage to trying to make the timing parts different.

That changes if you can't move the notches through enough frequency range with the control voltage schemes you have, so you may have to have different sections with different timing parts to get coverage of the part of the spectrum you want.

Actually, it's about this point in thinking that I decide to ditch allpass filters and go to voltage controlled notch and bandpass filters. The word "phaser" is catchy, and came from the phase shift nature of the way this was first done in electronic organs, but what one hears is the notches and peaks. If you're decoupling notch movement and peaks from moving synchronously, you may as well go directly to swept filters that do the job directly. One dual OTA like the LM13600 can give you a current-controlled notch with a 1000:1 sweep range, and also let you select notch, bandpass, highpass, and/or lowpass. Or you can do it the EH way with two sections of OTA doing an allpass notch and get the same range, about. But the more normal allpass stages with JFET or LDR phasing resistors works too. The range is just more limited.

@Dino and Jimi:
Actually, you guys already know the real inner "secret" of things sounding vocal. It's making two or more resonant peaks move around in that formant-frequency chart to hit the vowel-sound frequencies. Whatever makes them do that sounds "vocal".    There are more formants in human speech that the speech researchers recognize. The third and fourth formants play lesser roles, but do help with separating the sounds. Of course, I speculate that getting the wrong F3 and F4 might make it sound less vocal, so the problem of controlling three or four frequencies at the same time might comes up. That's just speculation. I've not tried more than two at a time yet.

[pinkjimiphoton]

R.G., did you ever build your vocal wah on the geo page? i've been looking at it, and thinking about trying a vero.

it would be nice to nail maybe 4 formants, using something like an old volume/pan pedal...the ones you could twist left and right for panning.

i'm still a long way off from being able to make much of the more complicated circuits, but your sing wah looks like it would be really cool...

i'd try and vero it (DINO??? help!!! lol) but don't wanna invest in something right now that i'm not sure would work...and it seems kinda open ended on the geo site, so i figured i'd ask.

on a side note, ordered another pot for my vocalizer, as it's crapping out...the conductive plastic pots aren't very robust, so my advice is solder them once...heat 'em more than that, and ya risk failure.

anyways. ...

[digi2t]

Just from my mucking about with 2 formants (i.e. Phase II, Vocalizer, Talking Pedal), I get the distinct feeling that for every formant you add, the effort (read: circuitry) required to sync it with the others must increase geometrically if you want to get anything interesting  Even at 2 formants, look at the Vocalizer (single vowel), and the Dipthonizer (multiple vowel). There's a lot of extra work there to get the extra vowels.

I think two is fine. After all, we have an expression in French that states; "Too much, is like not enough."

But... I did order 2 Dipthonizer boards from Madbean anyway. I'm curious to see what running them in stereo, each on different vowel settings, will yield.

God, I am completely shameless 

[Mark Hammer]

Well you're right about the advantages of using other approaches than allpass stages.

For those who really want things to be all vocally-like and formant-ish, consider that a simple single op-amp bandpass can be swept, just like a pair of allpass stages.  And if you mix the bandpass out of phase with the clean signal, you get a notch.  Sweep a couple of them in parallel, at staggered center-frequencies, and you have the functional equivalent of a phaser.  

Ironically, it can sound phaser-ish even if you don't mix them with clean out-of-phase.  I modded an Anderton Bi-Filter Follower ( http://hammer.ampage.org/files/Bi-Filter_Follower_redraw_with_mixing.gif ) to be able to stagger the two swept bandpass filters, and with certain "spreads" it sounds more like a phaser than filter.  I still have to do the invert mod, but Rick Lawrence (RickL) has done it and liked it a lot.  technically, a small addition of clean signal at the mixing node of an EHX Bass Balls would do the same thing.

But here's where it gets cool.  You could vary the resonance of each bandpass section independently according to its "role" in the sweep and position in the spectrum.  You'd still have to adjust the relative amplitude of each bandpass section at the mixing node so that it can sound like a phaser, but the character of what results could be made different than a conventional phaser.

[stringsthings]

I've noticed that phasers have "stages". What do more stages provide sound-wise. 4 stages, vs. 8 stages for instance? Is it "swooshier" ? ....

another control on many phase shifters is regeneration or feedback .... this takes the effected signal and sends part of it back thru a portion of the phase shifting circuitry ... generally, it makes the phaser sound more intense ... designer's have to be careful because too much regeneration can send the circuit into oscillation ...

the ratio of straight guitar signal to phase shifted signal is another possible control .... some phasers can sound like vibratos when you cancel all of the straight signal ...

[frequencycentral]

Phaser?

IMO more stages = swooshier. I've built many many phasers, most of the 'go to' DIYs, as well as designing a few of my own. I've built many 14 stage phasers and a handful of dual core 12 stage phasers.

The fun with phasing really beings when you consider the regen path - it doesn't just have to be over an odd number of stages (ie 3 like it the vast majority of phasers). if you've got 14 stages, you've got 12 choices of where to inject the regen path back into - each one reveals new aural territory. Then, if you think about a way to modulate exactly where the regan path goes - over odd or even stages........

http://dl.dropbox.com/u/967492/SDR%20Montage.mp3

Next up - phase cores in series and/or parallel, being swept independently, or by having one phase core swept by an inverted version of what the other phase core sees......you want formants, vowel sounds?

http://dl.dropbox.com/u/967492/Gemini/Gemini%20Montage.mp3

Haven't you guys been paying attention? I've been screwing around with this stuff for a while.......  


EDIT: Talking of phasers with synths, it's an old trick used by modular programmers to inject a subtle amount of white noise into the audio path along with the oscillators - just to give the phaser something to chew on across the spectrum. As for guitar use, phaser sounds so much more intense if you give it some dirt to chew on.

[stringsthings]

Haven't you guys been paying attention? I've been screwing around with this stuff for a while.......  

i still need to box up my Mark I  ....  ( i've been in fuzz box land for a long time .... )

[digi2t]

Haven't you guys been paying attention? I've been screwing around with this stuff for a while....... 

Rick, I'm a 47 year old millwright by profession, and I only got into this whole electronic/pedal building thing about 2 years ago. I still haven't hit that web page stating "You have now reached the end of the Internet... Please press "Back"!". 

Keep up the great work,
Dino

[Mark Hammer]

Oh, you mean the "404 at the end of the universe". 

[pinkjimiphoton]

i actually hit the end of the internet once, surfing around years ago. i kid ya not...was an actual website.

http://www.1112.net/lastpage.html

[Mark Hammer]

That is absolutely brilliant....and a wonderful idea.  I think I'll follow up.