JFET variable in resistor in Phase90 schematic

[bramm]

Hello all,

I'm trying to understand the schematic of the Phase90 guitar stompbox as described on https://www.electrosmash.com/mxr-phase90.
(direct link to the schematic here: https://www.electrosmash.com/images/tech/phase-90/mxr-phase-90-script-logo-schematic-parts-small.png)

I'd like to use the phase shifting part in a synthesizer module, with a control voltage input to control the phase shifting. (so I won't be using the original LFO, power supply or output/mixing stage)

I think I more or less understand how everything works, but there's something I don't get about the JFETS as voltage controlled resistors. As I see it:
+Vref is the virtual ground in this circuit
The LFO, connected to the JFET's gate, should oscillate between +Vref (=source) and +Vref-Vp (JFET pinchoff voltage). All clear until here.

But this is where I'm lost:
If I understand correctly, with JFET's as voltage controlled resistors, the drain is supposed to be higher than the source.
But in this circuit, the drain is an AC (audio) voltage, with the DC component equal to the source (+Vref), because of the AC coupling in the "input buffer" part.
So, half of the time the drain would be LOWER than the source...

Anyway, despite not understanding this part of the circuit, I went ahead and built the input buffer and phase shifting part on stripboard, using an external LFO (offset and scaled so it oscillates between +Vref/0V and -Vp) and a bipolar power supply (0V connected to +Vref), and what I'm seeing on the scope is a half wave rectified signal at the op-amps' positive input, and a distorted, DC shifted wave on the output.
If I replace the JFET with a potentiometer I get a nice clean phase shift, the way it's supposed to be.

I'd be very grateful if someone could point out what I'm missing here...

Thanks in advance!
Bram

[Andy Watson]

You are correct in your analysis. The FETs used in this circuit (J113 or similar ?) are almost symmetrical in that you can swap source and drain and their response is the same (give or take).
If you allow the audio signal to exceed -0.6V on the FETs, the gate-source junction will start to conduct and you will get a rectified signal - i.e. you need to think and use "small-signal" conditions to operate this circuit.

If you want to use more signal level without the distortion check-out the Morley Pro-phaser - it uses the same principles as the MXR but replaces the FETs with LDRs.
http://www.diystompboxes.com/pedals/prophaser.gif

[bramm]

Hey Andy, thanks for your reply!
Attenuating the input doesn't seem to change anything. I've tried with both J112 and 2N5457 (of which I have a matched quad).

One thing that puzzles me about JFET's being symmetrical is how source and drain (and thus Vgs) are defined.
Am I right that the lower voltage will become the source? And if so, shouldn't the audio stay above 0V instead of -0.6V?
(Sorry if these are stupid questions - it's the first time I use jfets)

+ thanks also for the link to the Morley Prophaser! I think I'll try it!

[Andy Watson]

Am I right that the lower voltage will become the source?

I think that is a reasonable assumption.

And if so, shouldn't the audio stay above 0V instead of -0.6V?

The audio is AC coupled into the drain/source of the FET so it should operate symmetrically about Vref. If the negative excursions are such that they fall 0.6V below the gate voltage (whatever that may be), the junction will turn-on.

[bramm]

It turns out that my JFETS have a very low VGS(off) of -0.5V.
I think that may be the problem here.
If I bypass the dc filter on the input and attenuate the audio to +/-0.05V I can get the phase shifting to work without distortion, but then noise becomes an issue.
I think I'll give up on this and try something with opto's or OTA's, or a lamp like the pro-phaser.

Thanks for all your help, Andy!