A simple way to get stereo in a phaser?

[Mark Hammer]

Here's a design idea for a fairly inexpensive and easy-to-make quasi-stereo phaser.  First some theory.

The location of the notches will always be the result of the sum total of all phase shift occurring in all stages.  Normally the location of maximal phase shift (i.e., what frequencies it is occurring at) is varied by means of FETs, photocells, or OTAs.  Some designs, however, will incorporate fixed phase shift stages in conjunction with the variable stages.  That is, the variable element (e.g., FET) is replaced with a fixed resistor and that phase shift stage is simply added in series to the others.  The MXR Phase 100 and Boss PH-2 are examples that readily comes to mind but there are many others.

In tandem with the cap, the fixed resistor produces additional phase shift that reaches its maximum (per stage) at a frequency dictated by the old standard F = 1/(2*pi*R*C).  When the resistor goes to ground and the cap to goes to the noninverting (+) input (i.e., configured as a highpass filter network), this introduces greater phase shift above the corner frequency.  When the cap goes to ground and the resistor goes to the + input (i.e., configured as a lowpass filter network), the phase shift is applied in the opposite direction (i.e., more phase shift below that corner frequency).  With a .022uf cap input and 22k fixed resistor to ground, a single stage would provide 90 degrees of phase shift for all content above 329hz (and lesser amounts of phase shift below that).  This would be on top of whatever the variable stages are providing in the way of cumulative phase shift at that frequency. 

As you might imagine, adding a single R-to-ground (I forget whether this is called "lead" or "lag") stage like that would result in more phase shift overall above the bass range and less once within the bass range.  If we flipped the position of cap and fixed resistor, we'd get the inverse: more phase shift as you moved into the bass range.

Okay, how does this get us stereo?  More particularly, how does it get us stereo with a conventional 4-stager like a Phase 90, Small Stone, etc.?  Let's say we fed the phase shifted signal to two output mixing stages.  Normally it would go to one, but since we want stereo we go for two stages.  These would be op-amp mixing stages not at all unlike the one used in the Ross Phaser (http://www.tonepad.com/getFile.asp?id=99).  Each stage combines the straight and phase-shifted signal in equal measure.

Here's the twist: the second output goes to two fixed phase shift stages (dual op-amp) and then to the output mixer.  So we have one output that consists of dry plus 4 stages of phase shift, and the other consists of dry plus 6 stages of phase shift.  The twist here is that we have one of the fixed stages lead and the other lag.  That is, one has cap to ground and the other has resistor to ground.  The component values are selected such that the two stages sum to produce 90 degrees phase shift across the entire spectrum, added to the sum of the variable stages. 

So, why bother?  What advantage is provided?  The result is that the notches produced by each mixed output are staggered relative to the other.  The stagger is produced without having to find more FETs that are matched, or power yet another LED/LDR combination off the same paltry 9v battery. The stereo produced is "more stereo" than what you get from simply having dry and wet coming out separate jacks.  Moreover, if you're into the sum vs difference thing, keep in mind that since there is more phase shift at one output than at the other (and it is evenly distributed across the spectrum), you won't get the cancellation in air effect that happens when you have two identical signals getting summed and differenced.

Whaddya think?  value for investment or more trouble than it's worth?  Seems to me that one could design a standard daughter-board with the extra 2 stages and a pair of op-amps for the mxiing stages, based on a quad chip like a TL074, that could be retrofitted to just about any phaser (assuming you have the space inside the box).

[Sir H C]

Most stereo phasers I know of just invert the LFO for left and right.  Your way makes more sense.

[Jaicen_solo]

Once again Mark, a simple but brilliant Idea. Have you tried it out yet?

[Mark Hammer]

Thanks for the compliment!

Haven't tried it out yet (wasted my first long weekend in a while refinishing furniture), but I hope sometime within the next month to put a rackmount 12-stage phaser together with the Ropez boards that Francisco Pena so kindly sent me.  That's not a 4-stager, as described in the first post, but it would be the same overall outcome of "staggered stereo".

While we're talking phaser mods here, folks should know that envelope modulation of the LFO speed (as in the old E-H Polyphase) is pretty easy to do.  In most instances, the LFO rate is governed by a single variable resistor.  Placing a photocell in parallel with that variable resistance and feeding the photocell with an envelope-driven LED should allow you to make the LFO speed increase in response to harder picking.  The amount of modulation would depend on how much impact the photocell has on the combined parallel resistance.  It is a trivial matter to stick a second variable resistance in series with the photocell so that you can adjust the minimum parallel resistance provided by the photocell+2nd pot.

Come to think of it, you can probably do that with just about any flanger or chorus too, couldn't you?

[notchboy]

I may be wrong here, but I don't think it's all that easy to get a fixed 90 degree phase shift at all frequencies.  Isn't that what the so-called "dome filter" used in frequency shifters does (approximating the Hilbert transform)?  That usually requires quite a complex multi-stage carefully trimmed circuit, not just two allpass stages.

btw, a similar idea of stereoizing a signal by running one channel through additional fixed allpass stages has been used in several fx, probably starting with the Instant Flanger.

[stm]

I put Mark's idea in the simulator and it works.  In fact, one possibility is having the two extra stages with interchanged cap/res locations, which indeed produces two notches more closely spaced than with the regular 4 identical stages.  The side effect is that both very low and very high frequencies exhibit a 6dB rollof, thus this arrangement may produce a "middy" second channel.  If the two extra fixed stages have the same phase shift polarity, then the second channel will have three notches which at some point appear interleaved with the two notches of the main channel.

I see potential for both combinations, thus a DPDT switch to reverse the order in one of the two extra phase shift stages might be a nice voicing addition for the second (stereo) channel.

Finally, I went one step ahead and tried mixing the main and secondary channels (right+left if you like), just as it would happen in a stereo to mono conversion.  In either case the depth of the notches is reduced, thus you don't get the theoretical mono phase shift composed of dry+wet channels.