Adding feedback to a Phase 90

[MoltenVoltage]

I just built a Phase 90 "script logo" clone and it sounds pretty damn lame compared to my MuTron and Boss Phasers.

The main thing it seems to be missing is a feedback loop.  Is there a way to add a feedback loop/control to it?

Thanks!

[frequencycentral]

http://www.geofex.com/FX_images/p180plus.gif

[MoltenVoltage]

Is that 47K pot in series with the 10K resistor the feedback loop?

I compared it with the Tonepad version:
http://www.tonepad.com/getFileInfo.asp?id=42

and they have a 22K resistor in red that appears to be part of the same feedback loop, assuming that's what it is.

Also, I have never seen a 47K pot, are they actually available?

Thanks!

[frequencycentral]

Yeah it's in series with the 10K. I built a 180 Plus, and had to increase the 10K to stop self oscillation with the pot maxed. I can get 47K pots easily in the UK. A 50K would do. Or even a 100K, with a well chosen resistor in parallel with it to reduce the resistance yet still prevent self oscillation when maxed out.

yeah R.G. replaced the stock 22k with a 10K plus the 47K pot.

[Mark Hammer]

The script version already has feedback, just not very much of it.  That helps it to sound pleasingly bubbly at faster speeds, but does take away much of the drama for long slow sweeps.

The basic premise of adding feedback is that it should never be above unity gain.  You have no idea a priori if any of the stages after the point of re-insertion add any gain.  On paper, a bunch of 10k inoput and 10k feedback resistors in each stage ought not to add gain, but with component tolerances, you could end up with the recirculated signal passing through 3 stages that unintentionally each have gain of 1.25.  This would mean that each iteration of the signal is amplified by a little under 2x.  In short order, you have runaway feedback/oscillation.  

Consequently, it is common practice to incorporate a trimpot in the regen/feedback loop.  The regen panel pot is set for max regen, and the trimpot is eased back to the point where the oscillation stops.  That way, the whole range of the panel control is usable.  You'll want something that lets you cover the range of something like 8k-20k.  That could be a fixed 6k8 resistor and 10k trimpot, a 20k trimpot on its own, or whatever you have handy to nail that feedback resistance that puts you juuuuuust at the edge of howl without going over.

Also worth considering:

a) Inserting extra stages of fixed allpass in the feedback path à la orange Ross Phaser.
b) Having less feedback for the bass than for the treble so that you don't sacrifice feedback energy/emphasis on the altar of oscillation prevention.  So, for example, less feedback resistance for the signal passing through a series cap to the return point than for the whole signal passing through a parallel resistor.  Since the treble tends to decay quickly anyways, there is less risk of oscillation incurring in the top regions of the spectrum.

[frequencycentral]

Also, it's all about well matched FETs of the correct Vgs, and adjusting the bias to get the deepest effect.

[MoltenVoltage]

Thanks for all the "feedback" 

I did some tests and found the following "ideal" setup for the 4 stage Phase 90:

Feedback loop:
B100K pot with 2 series resistors: 10K and 1.5K
10K alone makes it squeal when the pot is maxed.

Using 100K pot allows you to get back to the "stock" sound if you like.

Here is the BIG thing though:

RG's schematic has the feedback loop going into the FIRST stage, but it sounds MUCH better if you go to the SECOND stage like in the tonepad schematic.

[Mark Hammer]

That's actually an uncorrected error.  Most posted schematics for the P90 (and boy are there a lot) will show it going to the second stage.

Yep, the rule of thumb is that feedback should always be through an odd number of stages.  There are two exceptions to this.  One is the Small Stone, which feeds the signal back to the input (which I gather is inverting), and the other is the orange Ross phaser which sticks a fixed allpass stage in the feedback loop so that the regen signal always passes through 5 stages, even though the output is tapped from the 4th swept stage.

[MoltenVoltage]

Well, after YEARS of looking, I finally found the Mu-Tron Phasor II schematic!

http://www.experimentalistsanonymous.com/diy/Schematics/Phasers/Mu-tron%20Phaser%202.jpg

Anyway, it looks like the feedback loop on that goes back to the input stage rather than one of the phase stages (either that or it has an odd number of stages to begin with).  There is also an R/C combo to ground. 

This is consistent with what Mark said about an odd number of stages.  The Mu-Tron is the best phaser I have ever heard.  I haven't had time to really figure out what else is hidden in the schematic as its damn near impossible to read.

The LFO interfaces with a photocell.  It also uses some pretty odd voltages.

Just thinking out loud about where they hid the mojo generator...