phaser stages daughterboard?

[Mark Hammer]

If you look at the schematics for Rick Holt's "Causality 4 and Causality 6 phasers, he uses a pair of fixed allpass stages, but he sticks one before and one after the swept stages.  The Boss RPH-10 uses up to 4 fixed stages, which are shown here as IC6, two after the first set of 4 swept stages, and 2 more after the second set of 4 swept ones. 

The old orange Ross phaser, below, used a single fixed allpass stage in the feedback path; a rather counter-intuitive approach.  Even less expected, it used a lead, rather than lag, stage.  Huh?  If you look at the fixed stages for the RPH-10, they have a cap from the output of the previous stage to the non-inverting input, and a fixed (rather than swept) resistance to Vref.  This results in more phase shift at higher frequencies.  The Ross flips the R and C positions so that C goes to Vref/ground.  This results in more phase shift with decreasing frequency.

Here's where it gets really strange, though.  As you probably know from looking at P90 drawings, the feedback path always has to go back an odd number of stages.  Each phase shift stage is inverting, so travelling an odd number means the outcome is still opposite phase to how it started out.  Feeding back signal that is opposite in phase means greater cancellation, and hence notch depth.  You can control where in the spectrum this begins to happen by simply calculating the single-pole rolloff of the resistor and cap tied to the non-inverting pin.  With 10k and 47nf (I dare you to find a 50nf cap anywhere these days!), that increase or decrease in phase shift (depending on whether it's a lead or lag stage) as frequency decreases, begins around 338hz.

In a 4-stager, this would normally involve feeding the output of stage 4 back to the input of stage 2.  What the Ross does  (and remember, this is the orange one, not the black one that is a mostly copy of the Small Stone) is stick a 5th stage in the feedback path and feed that back to the input of the 1st stage.  So, it IS a 4-stage phaser, but uses feedback from a signal that has travelled 5 stages, similar to what you'd get in a 6-stage phaser.  And since it is a lead allpass stage, it produces more phase shift as it sweeps lower.  It seems to result in a more intense sound, judging by this comparison of 6 different phasers: https://www.youtube.com/watch?v=z4x6wWQNe3o

[pinkjimiphoton]

thanks bro!
i am printing the ross out and gonna look at it. i kinda like the way it looks, and i always loved the original ross phasers.

so i guess i'll add a single fixed stage as per the ross

use a footswitch to choose 45/90 and bypass the feedback circuit

a toggle to turn the feedback off or on

i think a pot would be good for feedback... thinking 50kish, but wondering if it  would be better to go maybe with a 25 k and a fixed resistance might be more stable?

appreciate the advice!!

[Mark Hammer]

The key thing is that you want to avoid the risk of oscillation from the feedback.  In principle, allpass/phase-shift stages are supposed to be unity-gain, but you can't really guarantee that, so there is a risk that, across however many stages the signal travels, there might be more than unity-gain in one or more stages.  If there was no feedback at all, that wouldn't matter so much.  The problem crops up when there is feedback.  If there is a cumulative gain of, say, 1.05x between the last6 stage and where the feedback path returns to, then that 1.05 gets multiplied by 1.05, and multiplied and multiplied, and pretty soon you're howling away.  The more phase-shift stages you have (4, 6, 8, 10+), the greater the risk of cumulative gain in the wet path.

Some pedalmakers just stick with a series resistance noticeably greater than what's feeding the inverting input from the previous stage.  So, the Phase90 feeds the inverting pin with 10k, and uses either 22k or 24k (depending upon issue) in the feedback path.  That assures that , even if there isz a teensy bit of gain from stage 2 to stage 4, there isn't enough feedback to produce oscillation.

Other pedal makers will use trimmers to get the most possible feedback prior to oscillation.  So, the during calibration, the manufacturer twiddles the trimmer for optimum FET bias, then sets the feedback/resonance to max, and adjusts a second trimmer until any oscillation just disappears.  That way, the end-user can know that no matter how high they set the feedback control, it won't start howling.  A number of flanger-manufacturers do the same thing.

[pinkjimiphoton]

so make r26 on the ross board say, a 250k trimmer, as its specc'd at 120k? or make the 10k resistor in the feedback loop of the added feedback ic the variable resistor? seems that would make the most sense to control oscillation, right? or should i make it a 25k trimmer? sorry for all the dumb questions. getting ready to do the vero layout for it, gonna probably go with ne5532's for the buffers, tlo74 for the quad and a 5534 for the feedback, as i dunno if i have any tl072's kicking around.

looking at the board layout from barry and comparing the two schematics it will be easy to graft the whole feedback circuit from the ross to the dr. phybes board, i can use the pads for the feedback resistor to attach the daughterboard.

i'm gonna socket the fets and chips so i can mess with them if need arises.

check back shortly. thanks again mark!

[Mark Hammer]

The 120k sets the absolute maximum feedback to be mixed in, and the Recycle pot sets amounts less than that maximum.  To wrap your head around it, think of U1A as a standard mixer with multiple signals feeding the inverting pin.  Normally, mixing in equal proportions would have all resistors feeding the inverting pin be equal value.  Since R3 =100k and R26=120k, such a mixer would "see" more signal via R3 than via R26.

In the Phase 90, the same thing applies, except that one input to the feedback return stage is a 10k resistance and the feedback resistor is 22k.  One can actually decrease the value of that 22k resistor to get more feedback in a P90, but I wouldn't risk much below 15k, and would also recommend sticking something around 33-39nf in series with that 15k to roll off bass in the feedback as further howl-prevention.

I will draw your attention to C11 in the Ross schematic.  Not only is feedback over many stages a risk for oscillation.  It is also a risk for noise.  Each individual phase-shift stage may geneate only the teensiest bit of hiss, but once you feed that hiss into a feedback loop - especially if there is even a tiny bit more than unity gain across all those stages - what stars out as imperceptible hiss can turn into truly objectionable hiss.  So, that added fixed phase-shift stage in the feedback loop includes C11 to roll off hiss in the feedback path, so that it doesn't get out of hand.  The rolloff doesn't start until around 16khz, so one can even afford to increase the value of C11 or its functional equivalent in some other circuit.

There is good reason to think about feedback/recirculation in a phaser as being somewhat like feedback in a flanger, however they are different in a very important way.  The feedback in a flanger is going to be up to maybe 10msec delayed from the input, so any summing of input and feedback isn't going to easily lapse into oscillation.  With a phaser, the feedback will be, at most, only a few cycles out of sync with the input.  The chances are pretty good that summing of input and feedback will lapse into oscillation, unless you do something to prevent it.

[pinkjimiphoton]

ok, so i will socket c 11. great tip!

i had a play with diylc, at first i was gonna use a dual, but it seemed wasteful... so i drew it as a single oa instead. don't mind the numbering, i didn't care about the numbers being right at all since its an add on board, but it should match the feedback loop of the ross ok




i'm gonna start with just this to see what happens.  can always socket r 26 too

feedback pot pin 3 to ground, i didn't add it to this graphic, b500k

[pinkjimiphoton]

boy this thing is turning into some serious spaghetti. holy shit. i hate to wire it up OUTSIDE a box cuz getting it INTO the box will require re-doing literally EVERYTHING
but of course, i haven't even fired the actual circuit yet, so i guess i gotta do that before i conscribe a box to its ultimate fate.

i have serious doubts about having to use that much wire inside a box like a phaser thats susceptible to a lot of noise.. wondering if i should abandon this endeavor lol





but kinda determined to see it thru

check out this horrible mess... wiring diagram i drew up, not showing the led's.

the 4pdt is set up to split the phaser from p45/p90 on rows a and b, row c switches the fixed phaser stage in or out of the circuit, and automatically shorts the input of the fb stage to ground in p45 mode, and row d will be green for 45, blue for 90. there will be a flashing speed indicator as well as the on/off led and the bicolor for the type. lotta wires. jesus. this is gonna have to all go on ONE pcb if it works

[Mark Hammer]

Yeah.....
This is pretty much why I like to avoid having multiple stompswitches, unless they are absolutely necessary for the pedal to do what it does.  If it's not the sort of thing one might use midsong, it's a toggle.  Case in point: order-flipping.  I think order-flipping is a very handy feature.  Can I think of a context where I'd need to accomplish it mid-tune?  Nope.

[pinkjimiphoton]

here's the thing with this one tho, a toggle isn't gonna actually make life any easier, still the same number of wires... and in this case, no where to stick that puppy!! it won't fit under the board in this case... the one for turning the fb on and off is actually ABOVE the board.

the bitch is, i hate re-working solder joints on pcb's. traces lift, pads disintegrate.. so doing it outside the box is not something worth doing. that said, tho, some of the other stuff i can't tinker with IN the box, which makes me suspect i may need to make a cardboard template to lay it all out on. but still then, gotta get it into the dang box afterwards.

AND no way to know if it will work without hooking it all up together. i made the daughterboard so it has a pin header to plug into a pin socket where the feedback resistor was... slick? nope, not too... cuz i'll have to undo the whole mess anyways to do the switching as envisioned. oy.....

so i guess i gotta start with the physical stuff of getting it all to fit.

drill holes for the footswitches.  make sure i get 'em right, cuz the whole mess depends on it.

drill holes for the  in/out jacks.. as close as possible to the stompswitches and positioned juuuuuust high enough i can run wires under them.,. and of course make sure i stay within the allotted area so i got room to juuuuuust fit the daughterboard...vertically, between the main pcb and the jacks.

drill out the holes for the speed indicator and the toggle, again within the confines allotted to them.

then figure out where to put the other two led's <thinking the 45/90 one a 3mm between the footswitches but it needs to be a 5mm so i can have it change color... yikes... but of course centered is gonna be hard to do cuz the 4pdt is physically wider than the 3pdt...

oy...... so basically gotta start with the grounds and led wiring and switches, jacks , etc first. if i'm lucky, this will all fit, but its gonna bust my stones, i know it!! lol

[Mark Hammer]

[pinkjimiphoton]

need a 125 instead of a sloop john b?

[pinkjimiphoton]

i have the actual phaser and daughterboard all wired up and ready to test as soon as i get done
workin'. gotta gig for another pedal company designing distortion/fuzz/overdrives, getting ready to have their second and third <mine> projects released soon commercially. kinda cool. i doubt it will be profitable

anyways... stay tuned, batman... more later...

will the phaser itself work?

will the daughterboard work?

if it does, will it howl its way into self immolation?

this and much much more will be revealed on the next episode...   

[pinkjimiphoton]

ok... got some stuff sorted.

don't ask me how long it took to notice i stuck tl072's in it instead of tl061's.



so it works. i didn't box it yet, as some circuit molestation is in order.
ended up with a tl074 for the phaser stages and 5534's for the rest.

it sounds remarkably like a phase 45 or 90.

i DID try the fixed phaser stage daughterboard posted earlier in the feedback loop. i tried it both in and out of phase actually. it does have an effect both ways.

its fairly subtle, i don't know if its worth keeping in this build. it tends to make it a bit easier to overdrive the whole mess.

that said, it does seem to work on p45 mode, a little bit. it sounds nice on p90 mode, but the knob doesn't really do a whole lot. i tried the trimmer all over and it didn't really make much difference other than in 45 mode, where it could make it oscillate a bit at full ccw on the depth pot.
also tried a plain pot in there for the feedback. not much difference between that and the ross snippet.

so... there are some good things

it DOES make the phaser more phasey
it really improves the "clarity" of the phaser... less muddy
give a slight "boost" to the phaser signal
works in both modes, to a point.. in 90 mode, works about as expected, in 45 mode mostly adds clarity to the upper end.

but the bad things
tons of spaghetti to wire
clicks or oscillates at some extreme settings of the feedback pot
doesn't really add all that much more to the original in term of phasing intensity
makes the circuit clip easier, and hard when it does.

i think in proper context its worth exploring, its also worth it if ya just gotta have "a little bit more"... i suspect i might have screwed this up somewhere a bit, as the feedback depth pot doesn't seem to do a whole lot, at least not as much as i expected. may need to play around with them resistors on the daughterboard a little, i think its restricting it a bit too much...

mark said you turn the feedback full, and then turn the trimmer up til it squeals and back it off i think... i don't get enough feedback thru it to actually make it squeal.

and yes, it does sound ok on the 2 stage phaser. makes it nice and clear, but... worth dealing with tonnes of spaghetti? maybe as a prototype for something where everything is all on one board.

i have to think on this and try it thru a bigger amp to see if its worth dealing with. i think i may just keep the added stage and fix the regen on full and take advantage of the slightly clearer phasing on the 45 and the nice more pronounced 90.

gotta do some experimenting still to see how i'm gonna do this. as originally envisioned, just plain not worth it. too much hassle, too little return. it'll be easier just to add the fixed stage in all the time, add a feedback pot after that, and do a toggle switch to kill it. one foot switch. saves a couple components and will do what it needs to do.

the mad scientist shit will have to wait.

for now... lol... stay tuned

[pinkjimiphoton]

when all was said and done, i left it mostly stock. i DID do some mods to it, but nothing major.

i found a definite sweet spot for the feedback resistor at 15.1k. i went with a 20k instead. maybe should change out to a 25k multi turn trimmer if i make another one.

the added fixed phaser stage didn't make the cut this time, but it is gonna be revisited in the future.

basically, i changed the resistor limiting the speed of the speed pot from 4.7k to 1.2k, now it will do that crazy fast vibrato like a nomad reverberola.

also stole an idea kinda from rick holt's causality phaser, and used a 1meg pot for course and a 10k pot for fine speed control, kinda a weird cojoined pot with one side of each pot going to the board and their wipers tied together so its like a giant custom tapered thing. sorta.
you can go from hyper sloooooooooooooooooow to a decent vibe tempo with the course pot at max, and then the fine pot takes over from the max speed of the 1meg all the way up to insects in your chromostones ring-ish modulation that's almost "harmonized".

also modded it so that there's an rgb led for the status, either p45 or p90. green p90, blue p45.  i cut off the red lead. i had common anode so i took a pulse off the speed pot pin 3 connection on the board to the center pin of row d on the 4pdt. a 5.6k resistor to b+ and it not only changes color independent of status, but blinks at the rate of the lfo so ya always know the tempo and which effect you're gonna have when ya hit the bypass switch.

i used row c of the 4pdt to automatically cut the feedback off in p45 mode with the "bottom" and middle pins.

row a and b i did ray wilkie's mod for the 45/90. by putting it all on the 4pdt footswitch it switches all the stuff at once.

there's still a toggle to turn the feedback off and on.

the rest of the circuit is stock, as is the bypass footswitching. got it in a 1590b with 9mm pots with ease. nice compact phaser that will cover a LOT of ground, and do nice vibe as well.

and if ya cut a switching jack trs, like one of them ground-isolated plastic wahwah kinda jacks in between the pots, you can plug an expression pedal of 10-100k in there and control the speed of the phase. its hip. i didn't do it on this one cuz i wanted it to fit a 1590b for my live board. but ya could definitely get away with it in a 125.

nice project from barry and the guys at guitarpcb well done