Warbling phaser

[Luke51411]

I built a dr phybes from guitarpcb.com http://www.guitarpcb.com/PDF%20Files/Dr%20Phybes%20v2.pdf
Which I believe is basically just a phase 90??? The closest fet match VGs I got was -.168 -.165 -.158 -.174 I wanted to get closer but this seems the best I can do with over 40 2n5457. The effect works but it sounds more warbly vibey than phase. It gets pretty intense at the fastest level and the sweep is not really heard below half of the pot sweep. I didn't have the opamps he called for so I subbed the 74 with an 84 and the 61 with an lm 741. I'm wondering if maybe I got an r or c wrong in the lfo and it is not a symmetrical sweep? The opamp wouldn't effect the sweep would it? Does the trimmer set the shape of the sweep? Maybe it's just not a precise enough trimmer...

[Mark Hammer]

The trimmer can be part of it, but more than likely it is a matter of the LFO itself.  The sorts of simple LFOs we are accustomed to using will generally provide acceptable waveforms with the listed component values...but not always.  I've had numerous instances where the LFO produced aomething that was sort of triangular, or at least produced an outcome that was sort of triangular, but not triangular enough.  Sometimes the resulting sweep sounded soradic, like a 30% duty cycle with a recurring triangular atop a DC voltage.  Sometimes it would be slightly skewed such that it would rise and fall at seemingly different rates.

I imagine that the ideal LFO waveform can be coaxed out of the circuits with some finagling of the component values.  For instance, a pin may require some fixed bias applied to it, via a resistive divider, and the resistor values shown ought to be the calculated ones, but one of them actually needs to be a little lower (or higher) than shown to provide the needed bias voltage.

The result of these imperfect LFO waveforms is indeed a sweep that just doesn't sound right, and maybe even a little warbley.  Not "musical".

Alternatively, mismatched FETs in a FET-based phaser can also result in a less than perfect sweep.  The reason why we match the JFETs, and also why there is a trimpot on the circuit board, is that we want JFETs that change resistance in response to all voltage changes in the LFO.  If one or more JFETs reach their individual point of no-further-change while the others are still moving/changing, the result sounds rather unmusical.  I liken this to the "turnaround" that competition swimmers make in olympic pools.  If they make a smooth efficient touch of the poolside and turn around to head in the other direction, it is a thing of beauty, and the component of a winning swim.  Matching FETs generally results in smooth turnarounds at both ends of the sweep.

[GibsonGM]

Did you socket the FETs, Luke?  I'm not entirely sure (not a big FET user), but you might need to have a closer match than that. They're acting like resistors, and if they're not close, you don't get much phasing.   Someone will come along and let us know if the ones you have are "close enough", I'm sure.     When I've done this, I'd accept the .168, .165, but not the other two, if that means anything....

I'd also try to get my hands on the ICs that are called for, that way you can take them off the table as the cause...the 741 should be able to do the job in an oscillator, but compared to the '61 IS junk, and I bet it could have some effect on the sweep.

Check each and every part of that LFO (too bad you don't have an oscilloscope!), if you got an R or C wrong YES, it would mess things up.  Badly.

**Edit: Mark got here first!!  )    **

[Luke51411]

They are socketed and this is the second set I've tried. The first set was not as close of a match but it still sounds about the same. I have a set of three that ar .555, .556 but the next closest is .572... The joys of fets... I'm going to put in an order tonight maybe I'll add a few more... I'm skeptical that the issue is the fets since it does sweep quite a bit it just seems like it's not in the right shape. The envelope of the sound isn't what I was expecting. It almost sounds more like adding modulation to a delay than phase shift. One other thing... I didn't have 150k resistors so I combined 100k and 47k but that is still well within tolerance.

[Luke51411]

I have 20 or so j201 would those work if I find a closer match? They are actually smd solder onto an adapter board. I heard the smd fets are closer tolerance.

[GibsonGM]

Yeah, so we're back to the LFO. If you got something really out of whack there, of course it won't work right.    Have you tried hacking an LED into it, to see what's going on?   

If the FETs aren't matched (and a couple of yours ARE, at least), you'd still have some phasing, just 'hard to hear', low volume of the phase itself, I'd suspect.   Seems like you have a diff. problem going on.

Yes, the 100K and 47K resistors in series are fine. 

Can  you post a recording of what you're getting?

[armdnrdy]

It gets pretty intense at the fastest level and the sweep is not really heard below half of the pot sweep. Maybe it's just not a precise enough trimmer...

Your description sounds to me like you have phasing, but it is very weak.

The phase effect will usually sound more "intense" when the rate is increased.

You seem to think that you can't quite get to that strong phase sound while adjusting the trimmer.

Change the zener diode from the 5.1 volt that is called out to a 5.6 volt. If that still doesn't produce maximum phasing, go the other direction with a 4.7 volt zener.

It is very common to match JFETs to each other but the VGS attained are not necessarily optimum for the phaser circuit. You may have to adjust the zener to make the phaser circuit match the JFETs.

If your description is correct, and you are encountering a "lopsided" "warbley" effect, then check all of the component values associated with the LFO section and check for solder bridges or splashes in that area as well.

[Luke51411]

Yeah I'll try to make a recording when I get a chance. Where would I splice the led in at? I'm assuming somewhere around r23?
I have some zeners around I'll try swapping that out. Is there a typical good range for vgs in a phaser or the circuit just needs to be tailored to the jfets?

[Luke51411]

I'm not sure if it's lopsided, it just doesn't sound like a smooth phase sweep. In my perception there is a range of frequencies it sweeps through faster on both the up and down sweep.

[Luke51411]

I think my phase 45 performs similarly though less extreme. Neither of these effects are entirely bad sounds and I kind of like the phase 45 as is but I think it is more of a vib/tremolo sort of effect as is though its hard to tell with it being such a subtle phase shift as is. Maybe I'll try swapping out the fets in that circuit and see if what that does.

[armdnrdy]

Where would I splice the led in at?

I don't think an LED is going to tell you anything that you don't know already. The LFO seems to be working so...the LED will just flash on and off.

Is there a typical good range for vgs in a phaser or the circuit just needs to be tailored to the jfets?

There are countless posts by both Mark Hammer and R.G. on this topic that are explained much better than I could.
Use the search function.
With that being said, to get more than adequate phasing without purchasing 100s of JFETs and finding a matching set that matches the circuit, the circuit bias has to be tailored to the JFETs.
That is the purpose of the bias trimmer.

[GibsonGM]

Where would I splice the led in at?

I don't think an LED is going to tell you anything that you don't know already. The LFO seems to be working so...the LED will just flash on and off.

Is there a typical good range for vgs in a phaser or the circuit just needs to be tailored to the jfets?

There are countless posts by both Mark Hammer and R.G. on this topic that are explained much better than I could.
Use the search function.
With that being said, to get more than adequate phasing without purchasing 100s of JFETs and finding a matching set that matches the circuit, the circuit bias has to be tailored to the JFETs.
That is the purpose of the bias trimmer.

+1    Yes, an LED would just go on and off.  It would tell you the LFO is working, which you don't seem too sure of (?)   A meter would do the same, esp. an analog one.  You could look for it at the FET gates, that rail that connects them all.

Gotta provide some sounds, Luke!  That will help anyone reading this to help YOU, as they may know more about what those sounds actually mean...

What Larry is getting at is that for certain phasers, etc, there may be a RANGE of Vgs your FETs would be best matched at.   So your -.168 etc may not be in the correct range to 'do much'!  The trimmer SHOULD allow you to dial in the proper bias for their Vgs, but if they're way out it may not be able to.   In order to make the circuit behave better, you'd have to alter the value of the zener diode, again as Larry pointed out. 

First try to post a sound clip, ok?

[amptramp]

One good reason to use the Paul Nelson PWM phaser - nothing to match, at the cost of a little more complexity.  The waveforms for a PWM phaser are here:

http://www.diystompboxes.com/smfforum/index.php?topic=97279.0

Somewhere on the net there is the schematic of the Paul Nelson PWM phaser design and I know because I downloaded it once but do you think I can find the link?  Not at the moment and I don't feel right about copying it here since he has copyrights all over the schematic.

[duck_arse]

luke, if your ic's are socketed, try swapping what you have in the '061 spot with what's in the '071 spot. just to see if your sweep improves, the audio quality will suffer.

+1 on fiddle the zener voltage, maybe fiddle the 3M9 at R23.

[Luke51411]

Sounds like I have some socketing to do. Good point on swapping the mono opamps. I already have those socketed. You never know when you might need to steal one for another project. I have the specified parts on order now.

[Mark Hammer]

You have two nicely matched JFETs in the circuit already.  You will still have a functioning phase-shifter if you were to remove the "less-matched" JFETs and make those stages fixed.  The 22k resistors in parallel with the JFETs set the maximum resistance of the JFET/resistor combination, such that any changes in voltage fed to the gate of those JFETs will result in a lower combined parallel resistance.

So here's what you do.  Pull the less-matched JFETs, and replace the 22k resistors in those stages with a 15k resistor.  This will fix where the phase shift of those two stages reaches its maximum, and the other two stages will continue to sweep.  Fixed+swept stages are frequently used as a way to increase phasing effect without requiring more LDRs or matched FETs, etc. (e.g., MXR Phase 100, Causality 4 phaser, etc.).  If your LFO is fine, and the warbley thing is purely a matter of matching all 4 JFETs, then 2 fixed + 2 swept should reduce the warbleyness, and yield a nicer sweep.

If the source is something else, then this test will provide indirect evidence of that.  Heck, you may even like how it sounds and decide to leave it like that.  Note that you may have to readjust the trimpot.

[armdnrdy]

On the thought of adjusting the trimpot....

When you get the most effect/phasing when adjusting the trimpot, is the trimpot all the way maxed in one direction?

[Luke51411]

No it's about in the middle, there is definitely a sweet spot with the most depth of effect.

[armdnrdy]

Okay...then you can rule out the zener.

[mikro]

I have same or similar problem with a P90 clone - plus some distortion.
my jfet match @12Kohm: 1600, 1606, 1623, 1645 mV.

Larry is right it's not the zener.
my LFO doesn't look asymetrical on the scope even quite close to sinus and still sounds jumping or warbley(?).
R.G. explained that for better matching we need several points on the resistance-range of jfets to be matched. so that might be our problem.
they are not changing same amount same time.