Compared phase shift stages between Phase 90 and Roland Jet Phaser

[nordine]

Yesterday i was looking at the phase shift stage structure on the Roland Jet Phaser and the Phase 90 and found them to be essentially the same:



Above image is the Roland one. Now, i noticed that the Phase 90 stage lacks R65, C29 and R66. Does this affect the sound at all?

[Kipper4]

To do a better comparaison, you'd have to compere the different lfo's they use too.

Then breadboard a buffer and a phase stage or two and do some futzing.

[teemuk]

Those components introduce feedback that reduces signal distortion introduced by the FET.

Here's a link to Siliconix application note AN105 that discusses using FETs as (voltage controlled) variable resistors. Read forward from page 2.

http://www.vishay.com/docs/70598/70598.pdf

[nordine]

To do a better comparaison, you'd have to compere the different lfo's they use too.

Then breadboard a buffer and a phase stage or two and do some futzing.

yeah, thought about running some tests, but wanted the theoretical part first 

Those components introduce feedback that reduces signal distortion introduced by the FET.

[/quote]

thats really neat. would that mean that implementing such a feedback network in a phase 90 could prevent the regen to sound yucky and prone to overdrive?

[Mark Hammer]

In a word, no.

The sound quality of the feedback IS partly a product of whether it results in distortion, but also arises from a bunch of other factors.

• The implicit assumption is that all phase-shift stages have unity gain.  The 5% tolerance of input and feedback resistors in each stage can result in any given stage having a little more than unity gain, which can result in much more than unity gain when all those stages are combined.  More phase-shift stages (6 vs 4) increases that risk.
• Since low-frequency content has greater amplitude, if the sweep goes fairly low, AND there is lots of low-frequency content in the feedback path, that can overdrive the JFETs.

Normally, the biasing of the JFETs sets the range of sweep, and where in the spectrum the notches will be produced as the LFO sweeps the JFETs.  If the sweep goes way down low, AND there is no "shaving off" of the bass in the feedback path, there is a risk of distortion.  If the phase-shift stages display any gain (and even 1.05x can be a problem), that risk is increased.

My personal phaser sensai, Mike Irwin, told me that while the additional network around the JFET, that you illustrated, does reduce JFET distortion, what it does is essentially raise the threshold for onset of clipping, not eliminate it completely.  In his view, when that clipping begins, it comes on really ugly.  he told me he preferred to leave the network out.  yes you do get a risk of distortion, but the distortion increases gradually with input signal level, and does not come on suddenly when some threshold is reached.

Now, I have to qualify this by noting that Mike tends to operate in the realm of synths, rather than guitar pedals, and synth signal levels can be quite hot, increasing the risk of hitting that threshold.  Moreover, the dynamic range of synths (at least analog oscillators) may well be more predictable and compressed than guitars.  At the same time, I would imagine that feeding a phaser with the output of an overdrive pedal is also going to be quite hot, and probably not that much different from synth-levels (which will likely be attenuated prior to hitting the phase-shift stages then boosted again at the output).

So, does any recommendation come out of this?  Yes.  I would suggest experimenting with the value of any caps  in the feedback path, should you experience any distortion.  Simply making sure there isn't too much bass (and especially bass with gain) hitting the FETs may be all you really need.  That's speculation, not gospel.

[nordine]

thank you Mark, that was really enlightening.

i noticed too that R66 is absent in Phase 90, instead it has a big resistor and a common modulation point that all jfets share. should it have any noticeable effect in how the phase is rendered? Jet Phaser one has all phase shift stages isolated from each other (at LFO point).

also, any ideas on what could be happening inside an Ace Tone Stereo Phasor? tried looking for schematics or even inside pics but didnt found any. only thing i could find was super short (like 4 second short) sound samples. i guessed it would be something similar to the Jet one, since Roland came from Ace Tone.