How do I implement regeneration/feedback to phase 90?

[vanessa]

Any help would be great!

[moosapotamus]

I think it already has (fixed) regen? the 22K connecting output of the last stage back to the inverting input of the first?

~ Charlie

[mdh]

I think the regen mod is discussed in terms of modifications to R28 (original MXR numbering?) here: http://diy.erikhansen.net/phase90.htm. On the Tonepad layout, this corresponds to the 22k resistor between pin 6 of IC2 and pin 7 of IC3, I believe.

FWIW, I ended up removing that resistor from my build, because doing so stopped the thing from ticking. I didn't really notice a loss in phase goodness, but I also didn't A/B it.

[vanessa]

I tried this pot mod and I did not notice that much of a difference. I guess I was looking for the type of regeneration that gets you more of the Small Stone "spacey" tone. I have a Morley Phaser One (MOD-POB), its regeneration control allows you to dial down to Phase 90 fuller phase tone, then dial up Small Stone "spacey tone" and in between. This is a really great phaser. It's just a little big (it also uses 2 9V batteries). I wanted to get the same versatility out of a Phase 90 and have the smaller footprint.

[Mark Hammer]

The distinctive element there is that the Morley takes the regen signal from AFTER the point where wet and dry are mixed, then feeds it back to the first phase shift stage, not the second, as is often done with FET-based phasers.  Of course, it can do this because there is an inverting stage in front of the first phase-shift stage.  This is yet another way of mimicking the feedback found in a unit with more stages, only using fewer phase shift stages to accomplish it.

The Phase 90 uses a default regen path resistance of 22k.  You need to consider the stage where the regen signal goes to as being like an inverting op-amp mixer (with some allpass duties on the side).  In an inverting op-amp mixer, the relative gain applied to signal A vs B vs C, etc., is a function of the input resistance.  When A and B have the same input resistance, then the same amount of gain is applied to them.  As well,  in an inverting op-amp, when input and feedback resistance are equivalent, you have unity gain.

Consider the regen signal as simply being another signal source mixed with the input.  By having the input resistance of the regen signal be larger than that of the "real" input signal, the regen is mixed in at a lower level.  In this instance, a feedback resistance of 10k and input resistance of 22k, means the gain for the regen signal in that stage is a bit less than x0.5.  Decreasing the input resistance of the regen path will increase the gain of the regen signal.  You never want to add any gain to the regen signal or else you'll break out into oscillation.  The one exception might be if there is a desire to produce oscillation.  Generally, though, you never go above a gain of 1 for regen.  So, variable regen for the P90 could likely consist of a 10k fixed resistor in series with a 50k or 100k variable resistance.  Since there is no DC-blocking cap between the output of the last phase shift stage and the return point, you probably want to build one in there for prudence' sake.  A value of .1uf seems about right.

[vanessa]

I'm going to try this. The Morley is like having two phasers in one box with that regeneration control (Phase 90, Small Stone). I've always liked the P90 over the Small Stone but would not mind both in a single unit.
BTW Mark, I tried your vibrato mod by lifting that 150k resistor on the dry output (that's R8 for you all). Very cool indeed! Thank you!

[vanessa]

So, variable regen for the P90 could likely consist of a 10k fixed resistor in series with a 50k or 100k variable resistance.  Since there is no DC-blocking cap between the output of the last phase shift stage and the return point, you probably want to build one in there for prudence' sake.  A value of .1uf seems about right.

Mark, from the mix output (back into the phase input) I would want to start with a .1uf cap (poly or electrolytic? If electrolytic what direction should it face?) then 10k fixed into 50/100k pot back into phase input?

Thank you!

[Mark Hammer]

So, variable regen for the P90 could likely consist of a 10k fixed resistor in series with a 50k or 100k variable resistance.  Since there is no DC-blocking cap between the output of the last phase shift stage and the return point, you probably want to build one in there for prudence' sake.  A value of .1uf seems about right.

Mark, from the mix output (back into the phase input) I would want to start with a .1uf cap (poly or electrolytic? If electrolytic what direction should it face?) then 10k fixed into 50/100k pot back into phase input?
Thank you!

Quite frankly, a ceramic .1uf is probably enough.  The rest of what you note is accurate.  Simplest thing to do is probably to remove the 22k from the PCB, and run two wires from the pads out to the pot.  Solder a 10k resistor to one outside lug and slip a piece of heat shrink over the resistor and lug to provide strain relief and avoid breakage (1/8w works best here).  Solder the .1uf cap to the wiper and cover with heat shrink if it will fit.

Note that because of the magical world of 5% resistors, using 10k as your minimum resistance DOES potentially place you at risk for oscillation at highest resonance values.  You could either choose to accept that risk and the possible auditory strangeness it might potentiate, OR you could use a 12k fixed resistor instead and play it safe.  You will note that many pedals using a variable feedback path often stick a trimpot in series with a fixed resistor to identify that point just below where the circuit breaks into oscillation at max resonance/regen.  You *could* do that, but what the heck, it's rock and roll...LIVE DANGEROUSLY!!

Finally, note that fairly high resonance settings can be more readily tolerated when the regen path has restricted bandwidth at the low end.  Remember that the bass is where the brunt of the signal lives.  If you chop the bass by means of smaller capacitor, you can live with more regen.  Anderton did that very thing in the Hyperflange, by implementing a 2-position bass-cut switch to facilitate more extreme regen settings/levels.  Though I haven't played with it, my gut sense is that restricting the bass in the regen path also makes higher regen settings more tolerable at faster speeds and creates more of a sense of "shimmer" than "wobble".  Indeed, restricted bandwidth in the entire wet path probably makes faster speeds more tolerable, but we'll stick to the regen path for now.

[vanessa]

Hmmm. I've tried a pot out from the 22k and it did not have the same effect as the Morley. The Morley regeneration adds an almost "bathroom reverb" effect to the outer sweep of the LFO. It's control acts like a depth control allowing you to dial in its intensity.

Looking at the tonepad schematic,

http://www.tonepad.com/getFile.asp?id=42

And taking your advice on how the Morley implements its regeneration, I figure that if you solder a trace just before the .05uf at the very output (this is the point where the two signals are mixed at) and run a wire back to the input of the first phase stage, say right where the 10k and .05uf meet. With a pot (100k?) in between controlling the intensity, I think I might get the same effect?

[Mark Hammer]

Hmmm. I've tried a pot out from the 22k and it did not have the same effect as the Morley. The Morley regeneration adds an almost "bathroom reverb" effect to the outer sweep of the LFO. It's control acts like a depth control allowing you to dial in its intensity.

Looking at the tonepad schematic,

http://www.tonepad.com/getFile.asp?id=42

And taking your advice on how the Morley implements its regeneration, I figure that if you solder a trace just before the .05uf at the very output (this is the point where the two signals are mixed at) and run a wire back to the input of the first phase stage, say right where the 10k and .05uf meet. With a pot (100k?) in between controlling the intensity, I think I might get the same effect?

Depends on the phase relationship.  Looking at the Morley schem, you can see that the phaseshift/wet path goes through 5 inverting stages, meaning that it comes out inverted, relative to its input.  The dry path, meanwhile, goes through a unity-gain inverting stage before it goes to the mixer.  So, the dry and wet are in phase with each other when they hit the mixing stage.  The mixing stage, meanwhile, is inverting.  This means that what gets fed back to the start of the phase-shift path is inverted.  In the case of the Morley pedal, the effect output is opposite phase to the input and bypassed signal.  Not good practice, but we'll live.

Whew!! You following that?  I'll give you a second to take a drink and get back on your bicycle.

So, to mimic this, do you need to invert the mixed signal before you feed it back to the start of the phase-shift chain?  I gotta tell you, I am so tze-mished by this schematic ( ), I'm not sure.  My gut reaction is that what needs to be done to truly mimic the mixed-down feedback signal conditions of the Morley requires more than what can fit on the Tonepad PCB.  Probably another op-amp stage is called for....but I'm not sure.  Too many inversions in there for me to think straight.