ROSS Phaser [again]

[lars-musik]

Hi fellows,

I am in the process of building a Ross Phaser and fail to implement 4 extra stages, my guess would be that the culprit is my lack of understanding where, what and why the "REGEN" net has to be connected (or disconnected) to or from.
Here are the schematics, the upper is the basic circuit the lower the extra stages I'd like to add.
The "REGEN" with the question  mark is unconnected at the moment.

What to do, dear group?

[Mark Hammer]

1) On mine (and many thanks to Francisco for sending me one of the expansion boards), I found that adding the 4 stages increased the risk of oscillation, especially since the additional stages bumped the range of sweep lower.  That is, the width of the sweep was the same, but where it started in the spectrum was lower.  That also meant that any peaks created around the notches, as one increased regeneration/feedback were also produced where the signal amplitude was already greater (in the low end), and I would get howling.  The solution was to trim back on the back in the feedback loop.  The schematic shows 1uf electrolytic caps at the input and output of the Feedback pot.  I changed them to 100-220nf and that fixed things.

2) Feedback in a phaser can be achieved in two major ways.  One is to have the phase-shifted signal recirculate through an ODD number of stages.  In a 4-stager, this is generally done by returning the output of the LAST stage to the input of the 2ND stage.  BY this means the phase-shifted signal passes through stages 2, 3, and 4 again - an odd number of stages.  I suppose the same thing could be achieved by feeding stage 3 back to the input of stage 1.  For a 6 or 8 stager, use your imagination.  There are many permutations and combinations that recirculate the signal through an odd number of stages.

The orange Ross phaser (FET-based) uses an interesting approach whereby it employs 4 swept phase shift stages, takes the output of the 4th stage, and feeds that back through a FIFTH unswept stage that is in the feedback loop, effectively mimicking the feedback sound of a 6-stager, while not using 6 stages.  (One of these days I have to try sticking such a 5th fixed stage in the feedback loop of a Phase 90.)

The other way of introducing feedback is to tap the signal at the output of the *mixing* stage, AFTER the wet and dry signals have been combined, and feed that back to the input stage.  In this instance, it would go from pin 7 of IC1b to pin 3 of IC1a.  The appropriateness of that approach would depend on a) having an active mixing stage  (which the Ross has, but many other phasers don't), and b) whether one or both of the input and mixer stages is inverting.

3) But to your question.  The added stages go where the diagram indicates - between points N and O.  The Regen can come from pin 9 of IC3b, and return to pin 13 of IC2b, as indicated.  Do note, however, that the more stages a feedback signal passes through (seven, in this instance, when the four stages are added), the greater the risk of oscillation, since even a smidgen of gain in any one or more of the phase-shift stages gets magnified via feedback/regeneration.    It's not a promise, but you may find that you need to increase the 27k resistor on the output of the Regen pot to, say, 33k, such that the entire range of that control is usable, without launching into howling.

[lars-musik]

Thank you very much, Mark! That opens a lot of room for experiments.
Your suggested modifikation to the 27K resistor (I guess you mean the one linling Pin2 and 3 on the pot, thus changing the taper?) didn't change anything.

I additionally replaced the 27K resistor between the 1uF and the REGEN net following the pot for a 50K trimmer (because maybe the regeneration feedback hast to be tamed a bit and that was the resistor you meant in the first place) and that makes the stock built without the extra stages much more usable - before that the effect was a bit subtle for my taste.

However, the extra stages still do not work. The sound created is very choppy and can be topped by a nasty oscillation with the feedback pot.  If othre people got this to work I must have a bug in there somewhere.