Phaser sweep implodes at bottom

[armdnrdy]

I'm working on a phaser project and everything's good except when the sweep gets to the bottom it makes some erratic sounds (kind of cuts in and out) before going back up to the top. I've tried four different sets and two different ways of matching the FETs. (matching VGs and matching VPs) all with the same result.

It almost seems as if the 2SK30s are shutting off at the bootom of the sweep. I know that Jfet VGs can vary quite a bit from different batches. Maybe the VPs are too high for this design?

Any thoughts?


Thanks,
Larry

[digi2t]

It would be nice if you post which circuit you`re working on.

I have learned that good phaser design depends greatly on the jfet`s being matched. When I built my Mr. Multi, I went with the first four I put my hands on, and it was alright (lucky). But, after doing some research, I sorted through them, and matched four of them using R.G.`s method on Geofex. They difference was night and day. Not even close.

I believe I used 2SK30`s in my Mr. Multi as well. I recommend that you start by matching your jfet`s first.

Cheers,
Dino

[armdnrdy]

digi2t,

If you read my post, you'll see that I did match the FETs, used four different closely matched sets and two different ways to match them.

I've adjusted and readjusted the bias to get a great sounding, full sweep but it has the issue at the bottom where it fall apart before returning to the top.


I'm wondering if anyone else has had this issue. Could it be because of the parameters of this batch of FETs? mThe VGs all measured below 1 volt.

[Mark Hammer]

I'm assuming that the FETs are biased via some sort of trimpot?  If that's the case, there will be a range of adjustment that still produces audible sweep, albeit in a slightly different range.  I'm wonderng if you've adjusted the bias "wrong".  Does it do the implosion at the bottom of the sweep even if you've biased them to sweep in a slightly higher range?

The other thing to consider is that jfets are generally used in conjunction with a fixed resistor to ground (or Vref) in parallel with the drain-source pins.  That resistance is part of what sets the frequency range where the sweep will happen.  If this is an unspecified "experiment" rather than any specific pre-designed circuit, perhaps the resistors used are suitable for a different set of JFETs to produce the desired sweep but NOT the K30s.

[armdnrdy]

Hey Mark,

It's the Jet Phaser circuit. I've twiddled the bias trimpot to find the "best" sounding sweep similar to the sound samples that I've heard. I'm using the 2SK30Y as specified in the factory drawings and visually verified in gut shots.

I have biased the sweep up and the issue goes away but at a heavy cost to the overall sound. ( It doesn't sound like a Jet Phaser anymore)

[Mark Hammer]

Each pf those FETs has a 330k/10nf network between gate and drain.  That network provides some degree of immunity to distortion on the part of the FETs.  Some years back, my personal phaser guru MIke Irwin told me that the drain-gate network did not, in his experience, eliminate distortion, but rather moved the threshold for onset of clipping higher.  Trouble was, he noted, that while a "non-immunized" FET-based phaser would distort at lower input signal levels, the onset of distortion was gradual, such that it got progressively higher at higher signal levels, the immunized FET would hold off and hold off, and when the signal exceeded the critical level, the distortion would come on suddenly and the distortion would be particularly ugly.

Now, given that Mike's specialty is in synths, where the average signal IS much higher than guitar, my thinking was that it was sound advice, but perhaps did not pertain to the guitar context quite as much.  But maybe it is more relevant than I think.

How could the effects of this sort of phenomenon show themselves only at the low point in sweep?  Well, consider that the highest amplitude portion of the signal is in the bottom end.  When any feedback is introduced, any peakiness around the notches produced results in a higher amplitude signal there.  The AP-7 does not have any sort of crude limiting in the feedback loop such as one tends to see in flangers like the BF-2, so the summing of feedback and input is free to range up as high as the supply voltage and gain structure allows.  That can result in the FETs receiving a higher amplitude signal when the sweep (and resulting peaks) that are in the feedback are at the low end.

So here is an experiment to try.  First off, with the bias set back to the "problem" range, kill the feedback signal.  The input and feedback signals are mixed via a 100k and 10k resistor (respectively) at the op-amp mixing node (there is no op-amp output stage; mixing of wet and dry is done passively in the AP-7).  Lift either end of that 10k resistor so that stage gets only dry signal.  If that cures the problem then we need a strategy to tame the feedback because, c'mon, how can you have a proper phaser without feedback?

Just ahead of the Feedback pot you'll see a 10uf cap.  Perhaps this might be a bit high and noit trim back on the bass peaks enough.  Consider droping its value to something under 1uf.

The other thing of note is that the grainy Roland pdf schematic shows a 330k resistor in the feedback loop of the very last phase-shift stage.  Since this adds gain, it places the feedback signal at risk of exceeding the threshold of "ugly clipping" for the FETs.  Maybe drop that down to 220k or perhaps even make it uniform with the other stages at 100k.