[sdiy] Stereo Phasing With the 32 Stage MultiPhase Project

[Scott Stites]

Hey René,

>On some of your samples there seemed to be an endless train of
>notches/peaks coming one after another. Almost like a barberpole.
>Thats where 6 stage phasers will fail miserably. Btw, are all those 
>stages at the same center frequency or are they offset (or even 
>offsetable) against each other?

Theoretically they're all at the same frequency.  In actuality, no way as far as I can tell.  

Dealing with this many Vactrols certainly drives home the point that Vactrols are by no means an accurate way of doing anything.  In fact, I've gone through a sort of limited Vactrol selection process.  I measured the least amount of resistance each Vactrol had in circuit and sorted out those that most closely matched.  Great - that leaves the rest of the curve =-D.  After that, I just did things by ear - I'd listen to four stage sections by themselves and mix and match (only a little bit) so that they sounded fairly close with the same sweep from four stage section to four stage section.  In the first section, I actually measured the percentage of how far off an LDR would be from the target average value and then tried compensating that by matching caps the same percentage in the opposite direction.  If that made a difference, it must have been very slight.  Anyway, if you listen to any of the stereo samples, you'll notice that the sweeps never quite sound the same on one c!
 hannel as they do the other, though I've gotten a lot closer than I thought I could, given the variance among the Vactrols. There's enough of a difference that, with identical settings on each channel, there will still be a stereo effect even if the channels are not modulated antiphase.  

I only have two control lines - one for each 16 stage section.  Adding more, as tempting as it sounds, would put more overhead on the panel.

In essence, I think a lot of the particular sound this phase shifter has comes from the variance (limited to some degree by a half-assed selection process) amongst the Vactrols.  Things stay pretty tight in the lower end of the resistance (upper part of the sweep).  Towards the bottom of the sweep, where I think the Vactrol resistances start to diverge more, there exists a whole slew of peaks and notches that go driving by on the sweep.  It's a fascinating, rubbery, albeit hard-to-control kind of sound down there I've not yet exploited for a sample.  I think this is one area where an envelope generator would really pay off.

>I know that's pretty big for a phase 
>> shifter, but this is something I can see using a lot (not just for 
>> phase shifting - the control signals will be useful, too).
>
>Right. They should also be brought out, to be used elsewhere. Or for 
>syncing etc...

Exactly - one can never have too many LFO's, I think.  

>> The phase and regen stages can be selected with rotary switches, 
>> though I have been considering using DGXXX type audio switch IC's 
>> with momentary switch selection.  
>
>If you ask me the stylisticly correct way of doing the switching in this 
>would be to also use vactrols. As I saw on your images it you have some 
>left over. :-)

=-D  I've accumulated these Vactrols bit by bit over a couple of years.  I added up how much the VTL5C3/2's would cost at one shot, and it was something like $60.  That would require drastic spouse-gating.  If I didn't have these elements on hand already, I'd certainly go the 'roll your own' route with a bulk bag of LDR's and some LED's.

>> The real consumer of panel space will be the control section. Not 
>> only do I have the LFO/ExtCV/and EF controls to worry about, but I'm 
>> looking at having a set of master controls so that both phase 
>> sections can be operated from a single set of controls.  The fact 
>> that one would *not* want to invert master center voltage and 
>> hypertriangular LFO's while everything else is invertable complicates
>>  that a bit.
>
>Isn't that just a matter of whether you sum prior to inversion and or 
>hypertriangulation or after that? (Geez, a block diagram would really 
>help there.. :-))

I've got one nearly finished for the audio path - it just has a few elements I'm not sure I want to keep or leave out.  The CV - ooooh, not even started.  Right now I'm just using LFO's and a mixer from the modular.  I'm running a triangle LFO through a FET distortion circuit to get the hypertriangular wave.  Not sure if I want to stick with that or try some other method.  In any case, 'inverted' hypertriangular (keeping the timing anti-phase but preserving the waveform) will require an inverted copy of the original triangle and then distortion of that.  

I've also thought of adding a quadrature LFO, but I think I'll avoid that until I see what fits.

Cheers,
Scott




________________________________________
PeoplePC Online
A better way to Internet
http://www.peoplepc.com