Tonepad MXR Pahse 90. Not much mojo.

[g1]

Looking at his response in post #2, I got the impression post #14 was suggesting some other type of matching (other than for resistance parameter) was more important ??? Turn on/off voltage?

[Mark Hammer]

"Do I average them?" One of the funniest comments I've seen here in ages. Don't even care if it's pointed at me. Funny is funny!

How about this? I'll rank the JFET attributes from most to least important, on a 10-point scale. Maybe that will remove some of the seeming ambiguity.

- Vgs in a range that results in the circuit's LFO voltage/current being able to vary drain-source resistance at all points in the sweep (10/10)

- Vgs matched across JFETs, within a few percentage points (9/10)

- range of drain-source resistance values (3/10)

The stages do not have to produce the exact same amount of phase shift at the identical frequencies. The notches created at the mixing stage will depend on the aggregated phase shift over the stages. So, at some point X in the sweep cycle, stage 1 contributes 82 degrees of phase shift at frequency F, stage 2 contributes 86 degrees, and stages 3 and 4 each contribute the hoped for 90 degrees. That's 348 degrees of phase shift at frequency F, at that point in the sweep cycle. One notch is produced and another is very very close to cancellation at that point in time, with the second expected at any moment now, as the sweep continues.

When one considers how poorly matched LDRs are, and what exquisite phasers they make, it becomes clearer that close-tolerance matching for resistance is way down the list of necessary or desirable properties. A nice-to-have, but not a necessary or sufficient condition for pleasing phasing. What matters most of all is that the resistance of all control elements - be they LDRs or JFETs - continue to change through the sweep cycle.

Given that the limited frequency range of our ears, musical instrument speakers, and the instruments we play, makes notches created above 10khz or so moot, then JFET resistance matters sort of. Clearly adding phase shift in that range doesn't do anything useful. (This is why both phasers and flangers tend to sound not as dramatic as hoped for when applied to a guitar, as opposed to a broadband mixed-down multi-track source, and why there is a point of diminishing returns beyond 8 or so phase-shift stages.)

So, the JFET resistance should ideally be somewhere in a zone that - with the cap values employed - results in usable phase shift. But there is still a lot of room in there to meander, and adjustments can always be made to the cap value or parallel resistor, to nudge the phase-shift range of a given stage this way or that. As builders, we appreciate the convenience offered by fixing the values of everything else, and then identifying a set of JFETs to suit what we've put in place.

One of the reasons why some designers prefer using OTAs or CMOS switches (+ PWM) as control elements is because assuring responsiveness throughout the entire sweep cycle is easier to achieve. LDRs yield lowest noise and highest immunity to distortion, but are cumbersome with respect to layout. OTAs aren't as cumbersome but traditionally have some headroom limitations, which brings a bit of noise with it. JFETs are convenient for layout purposes, but also subject to distortion. They can be made to behave a bit better, using the drain-gate network shown here. Mike Irwin - my own personal phase-shifter sensai - tells me that this circuit provides distortion immunity up to a threshold, but sounds terrible above it. He prefers the gradual increase in distortion, without the drain-gate network. One thing I've never quite fully grappled with is what sort of threshold we're talking about. Mike works primarily in the realm of analog synthesis, where signal levels are quite a bit higher than what a guitar or bass feeds a pedal. So his view may or may not be relevant to the stompbox context.

[Sea Chief]

as for the volume drop issue, there's a good advice right on the tonepad page:

"The resistor to change the level is the 150k which connects between the collector and base of the PNP transistor: 120k to reduce gain, 180k to increase it."

that's the (black) 150k resistor on the schematic, on the transistor closest to output. I remember having to increase the gain on mine also.
Maybe the output volume depends on how close the FETs are matched, so you need a little tweak if they are not a perfect match (just my assumption, may be totally wrong )

Chaps I will let you carry on the technical stuff- its way over my head!

Frus, re the 150k.. when referring to the "120k to reduce gain/ 180k to increace": which one 120k or 180k would result in -less- of a volume drop once pedal is on?

I noticed a 68k in there (I made pedal years ago you see), which I changed to a 150k.. but I got seemingly even more of a vol drop, and nasty thump switching on. So its removed but Im not sure what to put in, orig 68k back or what.

[frus]

make sure that you're changing the right resistor, it's this one



higher value should give less of a volume drop (try 180k or 220k)

[Sea Chief]

make sure that you're changing the right resistor, it's this one
.. higher value should give less of a volume drop (try 180k or 220k)

Yup thanks, i did indeed replace this to a 150k (orig a 68k in- why Ive no idea) but more of a vol drop, oddly, and a nasty ThUmP on switching. Orig 68k back. I think I'll have to just buy some of those ebay 4x FETS.

gratefully- SC

[J M Fahey]

think I'll have to just buy some of those ebay 4x FETS.

And pray they are actually matched

FWIW a few days ago I read about somebody who bought an expensive quartet of matched NOS tubes, guaranteed to have been so by using a computerized Amplitron (or Amplitrex) device.

3 of them were all over the place and 1 was actually shorted

[Sea Chief]

Good point.. though as ebay theyd be incined to answer with some numbers I could post up to check with you guys here if I asked how well-matched, & theyd be inclined to send awell-matched 4 Id think, or else (neg fdbk etc etc). This is where ebay comes up trumps I think, or should do.

[Mark Hammer]

$10, these days, can buy you enough 2N5952s that you can usually derive several sets out of them.