DIY 12 stage phaser pedal

[britt-stinker]

Hey everybody

First of, I have a question:

What exactly is meant by stage? In the beginnig I just thought it meant octave, but that cant be right.

Know to the real qouestion:
Is there any DIY 12 stage phaser? Like for example the moog? I know that its probably to complex to clone. But is there any 12 stage phaser? I have never seen a scematic/pcb layout for one.

Thanks

[toneman]

GEO's "Technology of Phasers......" is a GOOD place to start.....
Google works good also........
stayphased

[idlefaction]

hey man

check out technology of phasers of geofex:
http://www.geofex.com/Article_Folders/phasers/phase.html

it has a great explanation of how a phaser works and what's meant by stages.

Once you have your head around the 'each pair of stages gives you 180 degrees phase shift' idea, you should be able to see from a phaser schem like the phase 90 how you could just add stages to get a 12-stage, and what the advantage would be.

It's worth noting that in most commercial phasers (not the univibe) the stages are all tuned to the same frequency!!  So basically they are trying to get away with not matching their FETs by just having a whole pile of them.

HTH!

[britt-stinker]

hey man

check out technology of phasers of geofex:
http://www.geofex.com/Article_Folders/phasers/phase.html

it has a great explanation of how a phaser works and what's meant by stages.

Once you have your head around the 'each pair of stages gives you 180 degrees phase shift' idea, you should be able to see from a phaser schem like the phase 90 how you could just add stages to get a 12-stage, and what the advantage would be.

It's worth noting that in most commercial phasers (not the univibe) the stages are all tuned to the same frequency!!  So basically they are trying to get away with not matching their FETs by just having a whole pile of them.

HTH!

I rerad that thing and It really doesn't say anything to me.

[Mark Hammer]

The most obvious case would bein something like the Phase 90, where each "stage" consists of an op-amp with a cap going to one input pin, a resistor to the other, and a FET from the input pin to ground, among other things.  Each of these "stages" (also known as an allpass filter), can produce up to 90 degrees of phase shift at different frequencies.  The amount of phase shift is distributed across the frequency spectrum so that some frequencies encounter more phase shift than others, depending on the joint value of the cap and the resistance to ground in that particular configuration.  Two such "stages" can produce 180 degrees of shift for some frequencies which, when mixed with a nonshifted signal, produces a cancellation.

In principle, you can just keep adding more and more phase shift stages with the following constraints:

1) you'll have to make your own board since most layouts are for 4-stage designs.
2) some noise will accumulate over large numbers of stages and you'll have to keep that under control, especially if you intend to use regeneration/feedbacl.
3) if you thought matching 4 FETs for smooth sweep was a tough task, matching a dozen would be simply herculean in  nature; best to go with OTAs if you want to aim for lotsa stages.

[DiyFreaque]

With that many stages, I personally would go the optocoupler route myself (less noise inherent in the method).  Six VTL5C3/2's would get you there.

Having said that, if you want to go for a little-spoken-of-here route, why not go to 14 stages and try the ARP Quadra phase shifter that Juergen Haible adapted as a stand-alone unit?  Schematic, PCB trace and component overlay can be found here:

http://home.debitel.net/user/jhaible/additional_schemos.html

If you're unfamiliar with the sound of the Quadra phase shifter, think of the shimmery phase shifting of the synth instrumental in the middle of Styx's "Come Sail Away".

Cool thing about it is that it's a transistor ladder phase shifter, which you don't see every day (JH used CA3086 transistor array IC's for it, why he didn't go for CA3046's I dunno - maybe wouldn't make a big difference).  Being a transistor ladder, it should produce a fair amount of noise, but the Quadra had a very cool compandor setup to take care of that, and the compandor itself actually contributes a great deal to the sound of the thing.  Germanium diode clipping in the regeneration loop ...mmmmmm.

For detailed info on that awesome compandor, read up on his compandor notes for the Stormtide Flanger - he used the same thing in that.  JH loves the sound of it, and this is coming from a guy that's literally cloned nearly every high point of the of the analog world.  The Stormtide is linked on the same page as given above.

Cheers,
Scott

[puretube]

btw, Scott: did you follow zero the hero`s BadStone build-thread?  :wink:

[DiyFreaque]

Guess I must have missed that one, Ton.  Badstone - is that one six stages?

One day driving home from work I was wondering what in the world I'd do with all the cheap VTL5C2's I'd gotten from BGMicro.  It dawned on me that I could make a phase shifter out of them.  When I got home, I made a simple four stage phaser from the seat of my pants, and the thing actually worked, and sounded pretty good to boot.  Later I put an expo sink on it, which made it much better, and started adding stages.  I got to 10 before I got sidetracked.  

I found that once you get beyond six stages, the law of diminishing returns starts to kick in.  IE, six stages sounds quite different than four stages, but eight stages doesn't sound quite as different as six stages did in comparison to the jump from four to six.  Ten stages makes less of an aural difference when jumping from eight stages.  So, the more stages you go up, the more you have to add to make a big difference soundwise.  I think Mike Irwin explained the reasons for that quite well in a thread somewhere on this forum.   I imagine my next logical jump would have been to go to 14 stages.  The Tau Pipe had something like 20 stages, if I'm not mistaken.


Cheerio,
Scott

[Mark Hammer]

The Bad Stone uses a 4009 hex invertor chip as a set of six matched voltage-controlled resistors.  The ETI Phaser project, posted around somewhere, uses a 4049 like this also.  In one of his more capricious and ambitious moments, Mike made a 24-stager using a quartet of 4049s.  The demo I heard sounded magnificent, BUT:

- As anticipated, and as Mike mentioned to me and here as well, the noise was highly problematic and required a clever solution.

- The audible effect of having more stages CAN sound a bit more flanger-like but you need to have a wide-bandwidth signal to hear it.  The sample Mike provided me was done with a synth-derived broadband noise signal.  My guess is that 12 stages will sound fine but isn't going to sound all that "magnificent" if you plan on using it with an undistorted guitar.  The dramatic effect of flangers vs phasers is the increase in the number of audible notches as the delay time produced by a flanger increases.  Not only do the notches start lower down in the spectrum, but they also increase in number.  In the case of phasers, the number of notches produced remains constant, so the only way you're going to hear as dramatic an increase in number of notches when using a phaser is if the high end extends high enough that you could hear, for instance, the number of notches produced by a 12-stager go from 1 to 6 as it sweeps downward.  To do that, the 5 notches above that have to be so high that you can't hear them.  In other words, because you can't change their actual number, your only choice is to reduce their apparent number by having a wide-bandwidth audio signal and sweeping the phaser way up.

Scott's advice is probably appropriate, although I would say the advantage that extra stages buys you is a broader choice of where to tap and feed the regeneration signal.  In this spirit, Ross used a clever scheme for one of their FET-based phasers in which they used 5 stages.  Normally, because of phase relationships of a different sort, it is advisable to feed the regen signal through an ODD number of stages.  Typically, in 4-stagers like the P90, the last phase-shift stage is fed back to the input of the 2nd stage (i.e., the loop recirculates through stages 2, 3, and 4 - an odd number).  Ross added a 5th stage which was NOT swept, and sent the feedback signal through that, so that between stages 1-4, and the added stage, the recirclated signal went through 5 stages.

I mention this because you can apply the same logic to a 6-stager as well, adding a 7th unswept stage and sending the feedback signal all the way back to stage 1.  Or, if you felt like it, you could stick in a rotary or toggle switch and select where you feed it back to (e.g., stages 1, 3, or 5).

Finally, in the same spirit, take a peek at the MXR100, shown over at Generalguitargadgets.  This beast has 10 stages of phase shift, however only 6 of them are swept.  Remember that each stage produces a varying amount of phase shift across the spectrum (phase shift within each stage increases over frequency), and that the notches produced from the whole thing depend on the entire amount of phase shift produced across all stages.  So, if 4 unswept stages produce 33 degrees of phase shift each at some frequency X (132 degrees altogether), then the swept stages don't have to produce much additional phase shift at that frequency to generate a notch.

Scott, Mike, Ton, Steve, JC, et al:  One generally sees unswept stages having the same cap values.  What comments can you offer with respect to the possible effect of having additional unswept stages with staggered cap values, as in the Univibe or similar designs?  Is there anything to be gained in terms of the *perceived* sweep by distributing the fixed phase shift a little more evenly?

[idlefaction]

I rerad that thing and It really doesn't say anything to me.

OK then, which bit did you first get stuck on? A phaser is not a simple thing to understand, but if you really want to get your head around it I'm (and no doubt others who've already posted on this thread) happy to help.

If on the other hand you're totally new to electronics and the GEO article really doesn't make any sense, you'd be best to start off with a simpler circuit - check:
http://diystompboxes.com/sboxforum/viewforum.php?f=6

The short of it is that each pair of phaser stages makes a notch using some clever electronics, and this notch is swept up and down by an LFO.  If you match the components closely in your pair, the notch will be deep and very noticeable.  If you don't match them at all then the notch will be shallow and hard to hear.

Stompbox quality phasers generally use multiple pairs of phaser stages with the notch in the same place.  They do not use matched parts as this is very time consuming.  By stacking the shallow notches on top of each other, they add up to a deeper notch.  So even though the MXR Phase 90 is a 4-stage phaser and the Phase 180 is a 6-stage (i think?), they are actually just simulating a well-matched two-stage phaser because they only have one notched frequency.

So your query about a 12-stage phaser has a simple answer and a complex one.  The Moog unit probably (I don't know) uses matched parts and is a 'proper' 12-stage phaser with 6 deep notches.  You *could* build a 12-stage MXR phaser by just adding a bunch of phase stages into the basic 4- or 6- stage design but it will just make a deeper and deeper notch and sound probably the same as a 6-stage.

If you would really like to build a DIY 12-stage phaser with 6 deep notches, let us know and one of us might design you one.  ;-)

HTH...

[Paul Perry (Frostwave)]

The other way to go for a zillion stage phaser is to use a switched resistor approach. Pretty easy to get matched resistors
But if I was buiding a 12 stage phaser, I wouldn't, I'd make two 6 stage phasers instead. Then you have a lot of flexibility (esp if you add an option of seperate LFOs!)

[puretube]

free information block:

[britt-stinker]

I rerad that thing and It really doesn't say anything to me.

OK then, which bit did you first get stuck on? A phaser is not a simple thing to understand, but if you really want to get your head around it I'm (and no doubt others who've already posted on this thread) happy to help.

If on the other hand you're totally new to electronics and the GEO article really doesn't make any sense, you'd be best to start off with a simpler circuit - check:
http://diystompboxes.com/sboxforum/viewforum.php?f=6

The short of it is that each pair of phaser stages makes a notch using some clever electronics, and this notch is swept up and down by an LFO.  If you match the components closely in your pair, the notch will be deep and very noticeable.  If you don't match them at all then the notch will be shallow and hard to hear.

Stompbox quality phasers generally use multiple pairs of phaser stages with the notch in the same place.  They do not use matched parts as this is very time consuming.  By stacking the shallow notches on top of each other, they add up to a deeper notch.  So even though the MXR Phase 90 is a 4-stage phaser and the Phase 180 is a 6-stage (i think?), they are actually just simulating a well-matched two-stage phaser because they only have one notched frequency.

So your query about a 12-stage phaser has a simple answer and a complex one.  The Moog unit probably (I don't know) uses matched parts and is a 'proper' 12-stage phaser with 6 deep notches.  You *could* build a 12-stage MXR phaser by just adding a bunch of phase stages into the basic 4- or 6- stage design but it will just make a deeper and deeper notch and sound probably the same as a 6-stage.

If you would really like to build a DIY 12-stage phaser with 6 deep notches, let us know and one of us might design you one.  ;-)

HTH...

I would like that.

[SeanCostello]

I found that once you get beyond six stages, the law of diminishing returns starts to kick in.  IE, six stages sounds quite different than four stages, but eight stages doesn't sound quite as different as six stages did in comparison to the jump from four to six.  

I think that, going up to 12 stages, you hear major differences with every increase by 2. Eight stages is not nearly as dramatic as 12 stages. Still, once you get past 12 stages, the change in sound is less dramatic...

...until you start getting past 64 stages or so. Once you do this, the dispersion of the allpass stages starts to have an audible effect. High frequencies start coming out earlier than lower frequencies, causing a "chirp" similar to spring reverbs. Sweeping the allpasses causes some pretty crazy effects with this many stages.

I have no idea how to do this in analog without creating a massive amount of noise. My experiments with this have been in digital. The phase shifting module I have running on the SHARC can easily compute 256 swept phase shift stages without choking the DSP, and that is unoptimized C code.

For standard guitar purposes, 4 to 6 stages sound best to my ears.

For reggae, a 12-stage phase shifter produces the classic dub sound. I believe Lee "Scratch" Perry used a 12-stage unit, although I forget the exact model.

For space music, my favorite topology is based on the Schulte Compact A phaser: 8 phase shift stages, but with negative feedback around the first 2 stages only (with a simple R/C lowpass filter in the feedback loop), causing a pronounced single resonant peak in the output signal. Used by Kraftwerk, Klaus Schulze, Tangerine Dream, and others.

Sean Costello

[StephenGiles]

I have never built beyond 12 stages, which was if I remember, 2 badstones in series and didn't sound that much more remarkable than a single one. What occurred to me was that a multi multi stage phaser which may sound good when you are playing alone at home, will most likely not sound that much different from a bog standard 4 stager in a band situation, and makes me wonder if it is really worth the time and effort ............................ on second thoughts of course it is!
Stephen

[idlefaction]

at first when i saw mention of a 64-stage phaser i initially thought 'isn't that approaching a flanger?' but then remembered that flangers have a set relationship between the notch frequencies and of course with a phaser you're free to put the lil buggers anywhere you please :-D

britt, I will have a go at a 12-stage phase design later today if time allows.

interesting re: lee scratch perry, the guy is a genius, i want one of those thingies  :-P
so what's a clever way of reducing noise on multi-stage phasers then?  companding comes to mind. Using Twin-T's instead of regular phasing to get a notch cuts your stages in half...

[puretube]

low impedances...

[DiyFreaque]

Ton - Spiffy!  Thanks!  I recognize the topology now (it was discussed on SDIY a while back IIRC).

Sean - you're right about the stages, I bumblingingly (how's that for a word?) was trying to say that things started to sound less dramatic (to my ears) when going from a number of stages to a higher number of stages, though it is dramatic enough.   In fact, I've played with it in C-sound, and for some reason, I'm thinking the file I started with might have originated from you(?).  It was originally an ensemble chorus through phase shifter setup IIRC.

The phase shifter you talk of with the negative feedback around the first couple of stages sounds really neat.

Mark - I don't have any experience at all with playing with fixed stages and modulated stages together (never thought about doing that), but it does sound interesting.  Sort of like the theta processor of JH's Stormtide Flanger?

I have heard Rene Schmitz's tube phaser, which he set up so the stages could be spread out, I guess kinda like a Univibe, and that was a remarkable sound....  That's one of the sweetest phasers I've ever heard, BTW.

Cheerio,
Scott

[SeanCostello]

at first when i saw mention of a 64-stage phaser i initially thought 'isn't that approaching a flanger?' but then remembered that flangers have a set relationship between the notch frequencies and of course with a phaser you're free to put the lil buggers anywhere you please :-D

If you set all of the 1st order phase shift stages to the same frequency, the effects starts to get closer and closer to delay-based flanging, with the exception of the dispersion effects. A 24-stage phase shifter, with all phase shift stages at the same frequency, sounds rather flangy.

With 1st order stages, it is hard to map out the notches exactly. 2nd order stages allow for precise placement of notch frequency and width, althogh notch depth will probably be the same for all notches if you have a bunch of 2nd order stages in series.

Does anyone know of a topology for a voltage controlled 2nd order allpass stage? In the digital world (where I spend almost all my time nowadays), 2nd order allpasses are easy to construct, and if you use the lattice form you can have nice properties even in fixed point. However, I don't recall seeing any analog 2nd order allpasses where you have control over frequency and Q. Maybe something could be constructed based around a state variable filter...

Sean Costello

[puretube]

"I don`t trust 2nd order allpasses..."  :shock: