Tweaking a JFET phaser

[mdh]

I have built a couple of JFET phasers (Phase 45 and 90), and while I like them, I always find myself wanting a little more.  But more of what?  I'm a little unclear on the limitations of these designs, and the benefits of tweaking different elements.  I was hoping that some of you folks might be able to help me get a better handle on this.

At the moment, I have a 6 stage version of the Phase 90 (Phase 135?) on the breadboard right now, and I have room to add two more stages pretty easily, and could probably fit another two beyond that.  Six stages give "more" phasing in some sense (more notches, is my understanding from a theoretical standpoint).  I guess I would describe the sound as more lush, maybe chorus-like.  I'm definitely going to try eight stages, but I'm not sure that more stages is the "more" that I'm looking for.

I think that maybe what I'm actually looking for has more to do with the LFO than the phase stages.  I think I'm looking for a more swirly sound, which may mean that I need a more sinusoidal and possibly deeper LFO.  I know R.G. has put up a couple of schematics of Phase 90 derivatives with different LFOs, but I've been thinking about trying to use something like the Small Clone LFO, since it has so much depth to it.  I'm wondering if it's reasonable to use something with so much depth, though, given that the voltage swing at the gates of the JFETs is only about 300mV (at least, that's what I'm measuring on my breadboard).  The square wave of the LFO is banging between about 1.4V and 7V, which makes me think that the attenuation before the LFO signal gets to the JFET gates (is this due to the 3.9M resistor?) is an intentional design feature, rather than an incidental limitation of the LFO.  Would a larger swing at the gates lead to distortion or something?

I think the sound that I'm looking for is kind of analagous to TZF in flangers, except maybe a little less dramatic.  That feeling where the phaser sweeps to an extreme and for a second you kind of think your head is turning inside out.  Could be that I'm asking too much of a phaser, but I've certainly heard digital models of phasers that sound like what I'm trying to describe.  Again, I'm not sure if this is more a matter of LFO depth or shape, or if it has anything at all to do with the number of stages.  Could it be that I'm just running into the limitations of JFET phasers?  Would I be better served by an OTA phaser such as the Small Stone or Ross Phaser, or by something that uses LDRs, like a Phase 100 or one of the Mutron phasers?

Any and all thoughts are most welcome!

[Mark Hammer]

It may well be that what you really want is simply a means to manually tune where the sweep range is so as to match it to the tune.  Sometimes, just about any swept effect can seem to be less majestic or whatever, because it "makes the tag" at the end of the sweep range and turns around before you were ready for it to do so.  Part of the legend of the A/DA flanger is the extremely wide sweep it is capable of.  It just keeps going and going.  I modded my Ross Phaser to have a slightly wider sweep and I have to say that when it keeps going after it should have stopped you really notice....and that's still only 4 stages.

Incidentally, the kerfuffle over matching FETs in a phaser is precisely because it permits equal intensity across the entire range of sweep.  When FETs are unmatched, users may encounter "FET fatigue" as individual FETs go "Nope.  That's it.  I ain't sweeping any farther than this.", such that you may have 4 stages of phase shift in the middle of the sweep but only 3 or maybe even only 2 as you get to each extreme of the sweep.

[mdh]

Incidentally, the kerfuffle over matching FETs in a phaser is precisely because it permits equal intensity across the entire range of sweep.  When FETs are unmatched, users may encounter "FET fatigue" as individual FETs go "Nope.  That's it.  I ain't sweeping any farther than this.", such that you may have 4 stages of phase shift in the middle of the sweep but only 3 or maybe even only 2 as you get to each extreme of the sweep.

Well that's interesting, because we only match FETs at one point now, don't we?  If they're so variable, do we have any guarantee that FETs with similar values of Vgs_off will have very similar responses away from that point?  I.e., that the graphs of drain-source conductance vs. gate voltage will all lie on top of each other for a set of "matched" FETs?

I have more to say to the first portion of your response, but I have to get ready for a meeting in < 1 hr.  Yeep!

[notchboy]

Letting your job take priority over posting here?  Bah!  You need to get a government job, like some people here.

Seriously, if you're worried about variability of sweep response, be glad you're not using Vactrols...

[Mark Hammer]

Letting your job take priority over posting here?  Bah!  You need to get a government job, like some people here.

Heh, heh....

Seriously, if you're worried about variability of sweep response, be glad you're not using Vactrols...

Ah, but there's the rub.  Vactrols may not be easy to match for resistance, but they can almost always be coerced into changing their resistance in response to more LFO signal.  Keep in mind that it is a change in effective resistance that results in the location (in the spectrum) of maximum phase shift changing; i.e., producing a sweep.  Even if the LDRs are 10-20% off from each other in terms of equivalent resistance, if all the LDRs change resistance value in response to that extra 250mv of LFO signal fed to the LEDs, then the notch location will change.  You will note that several of the more desirable phasers of all time, the Mutron Biphase and MXR Phase 100, both use LDRs.  OTAs will provide for more precise matching, and certainly lower current requirements, but LDRs afford nice clean performance.....and they don't HAVE to match.

Keep in mind that the goal is to arrange for a cumulative phase shift across the stages that will result in cancellations.  In a 4-stager, each stage can give you up to 90 degrees at some designated point.  When one or more of those 4 stages grinds to a halt, and the remainder keep sweeping (whether up or down) the total phase shift at frequency F is not likely to be the full 360 degrees.

[oldrocker]

I recently removed the univibe caps in my Easyvibe and put in all .01uf caps to get a stronger phase out of it.  It isn't exactly what I was looking for but it sounds ok.  What I was trying to get going was a feedback circuit.  I tried it but I can't seem to get it to work.   I don't know why.  I tried it back to the second stage, first stage you name it but nothing seems to work.  I used the Ross Phaser feedback design but that didn't work right at all.  Any ideas.

[mdh]

Letting your job take priority over posting here?  Bah!  You need to get a government job, like some people here.

Heh, heh....

Well, I do work for a state university... as a 7th year graduate student.  So there is some incentive to wrap things up!

It may well be that what you really want is simply a means to manually tune where the sweep range is so as to match it to the tune.  Sometimes, just about any swept effect can seem to be less majestic or whatever, because it "makes the tag" at the end of the sweep range and turns around before you were ready for it to do so.  Part of the legend of the A/DA flanger is the extremely wide sweep it is capable of.  It just keeps going and going.  I modded my Ross Phaser to have a slightly wider sweep and I have to say that when it keeps going after it should have stopped you really notice....and that's still only 4 stages.

After experimenting with added stages (up to eight), I'm pretty sure that what I really want to do is see if I can mod or replace the LFO to optimise for a long, grandiose effect.  If I'm understanding your terminology, part of this is sweep width, but part of it is also sweep depth.  I'm assuming that by "width" you mean the period of the LFO, and by depth I mean the amplitude of the LFO cycle.  It seems like the stock Phase 90 LFO can actually have pretty long periods, but the maximum depth isn't sufficient to get the sound I'm looking for.

The crux of my question is whether I can sweep the gate closer to the source voltage without experiencing unwanted side-effects (distortion, LFO noise?).  I've been playing with the value of the large resistor between the +ve end of the LFO integrating cap and the FET gates, and it seems like I can get more depth by decreasing that value.  I was getting something like 300mV with the stock value, and I seem to be getting 500-600mV with the value reduced from 3.9M to ~2M.  It's hard to tell if this is actually having the intended effect, because changing that value means that you have to rebias the FETs, and by the time that's done, I don't remember what it sounded like before.  I think I'd notice a dramatic change, though, and I'm definitely not there yet.

Any idea if I'm on the right track, or am I barking up the wrong tree?

[Shepherd]

Build the ropez.  You won't look back.

http://tonepad.com/getFileInfo.asp?id=99

OTA FTW

[nordine]

I was getting something like 300mV with the stock value, and I seem to be getting 500-600mV with the value reduced from 3.9M to ~2M.  It's hard to tell if this is actually having the intended effect, because changing that value means that you have to rebias the FETs, and by the time that's done, I don't remember what it sounded like before.  I think I'd notice a dramatic change, though, and I'm definitely not there yet.

Any idea if I'm on the right track, or am I barking up the wrong tree?

hi matt

well, thats what i did, sometime ago, for getting more "phase" of it ..i remember setting it on 1.2M  ..you know thats pretty much decrease... the feel of the wave 'squared', but i got what i was after ... you could record samples to compare the results

[mdh]

Build the ropez.  You won't look back.

Thanks, I just might do that.  It's been awhile since I've ordered from Steve, so maybe I'll do that soon and pick up the OTAs.  I've been buying my parts locally of late, and no one seems to carry OTAs.

hi matt

well, thats what i did, sometime ago, for getting more "phase" of it ..i remember setting it on 1.2M  ..you know thats pretty much decrease... the feel of the wave 'squared', but i got what i was after ... you could record samples to compare the results

Well, it's good to know someone else has tried this.  It's not really giving me the sound I'd like, though, and I have gone all the way down to 1M.  I think I'll play around with some other LFO designs, and see if I can figure out how to graft them in.  Thanks.