Phaser Feedback - OTA desing

[Epameinondask]

I've been trying to design my own phaser the couple of months, and so far so good. I'm using 6 and 8 stages and it's roughly based on the Ross Phaser with the LM13700. I've achieved a nice vintage sound without using feedback. But of course, I want do add the feedback function, but I ran into a "problem"

The sound and character that I'm trying to achieve with the feedback is that of the Bad Stone (reissue),which can go from very subtle to alien-like sound with the feedback turned fully on ,still maintain a very liquid sound in the fast rate settings and in the slower rate settings it has that swooshy jet fighter like sound.

I've been doing a lot of reading on phaser feedback from related threads of this forum, so I tried pretty much every combination that has been suggested or seen in famous and well known phaser schematics. The only sound I get from the classic capacitor-pot-limiting resistor configuration is a sound that has very little difference from the no-feedback-sound. In addition, I can hear in the background an almost oscillating sound, that of the oscillation you get if you do not use the limiting resistor, but still everything works "ok". No noise. Just that awful feedback.

The phase shifting stage is exactly like the Ross Phaser but with changed values. 27k is 100k, 1.8k is 1.5k, 33nf is 4.7nf , 10k is 47k and the resistor from the LFO is 8.2k. The feedback limiting resistor is 120k and the cap is 100nf and the trimmer I use is 500k.

To keep things simple, I use my function generator for the LFO duties. So, what am I doing wrong here? Someone would expect a very alien like sound coming from a 6 or an 8 stage phaser with feedback fully turned on. Any suggestions?

Thanks in advance.

[Epameinondask]

It seems that the feedback sound I'm after can only(?) be achieved by feeding the output back to the input of the circuit.The Bad stone does it this way, but it blends the signals passively if I'm not mistaken. By doing some LTSPICE simulations, I see that input and output wave forms are exactly out of phase, which makes perfect sense because the mixing opamp is inverting the signal.

So to achieve this kind of feedback I need to invert the output signal, right? Could a simple common emitter NPN do the trick? How should the feedback configuration be? In a (maybe ignorant) attempt to use a NPN LPB1 style at gain 1, I only made the phasing worse. Is the use of an inverting opamp maybe better suited to this feedback function?

[Kipper4]

Check this out.


http://www.diystompboxes.com/smfforum/index.php?topic=69366.msg556197#msg556197

[Mark Hammer]

Generally, the more phase shift stages - of any variety - the greater the need for some sort of fine-tuning of the feedback, since the risk of oscillation increases with the number of stages.  The stages are supposed to be unity gain, but if stage 1 is 1.02x, stage 2 is 1.025x, etc., it adds up over all those stages to more than unity gain, at which point oscillation occurs.

There are two cures:

• use a smaller value of cap in the feedback loop so that obnoxious low-end is attenuated to prevent howling
• use a trimmer to set maximum feedback before oscillation sets in

[Epameinondask]

Thank you all for your input.

Kipper4, I've already read that thread, but it doesn't seem to give any info on how to configure the feedback path.

If maximum feedback effect is achieved from output back to input, then in my case I need to invert the signal. But I've already tried that with no success.

There are two cures:

    use a smaller value of cap in the feedback loop so that obnoxious low-end is attenuated to prevent howling
    use a trimmer to set maximum feedback before oscillation sets in

Well, I'm aware of how to not have the feedback self oscillate. But I haven't tried smaller cap values. Thanks! But the thing is that I would like to achieve the Bad Stone type of over the top feedback. It seems that feedback from the output back to the input is the way to go. But as I mentioned before I tried to invert the mixed output signal and then feed it back to the input with no success.If my assumption is correct, what does the inverted feedback signal have to look like in order to achieve that Bad Stone style feedback? Does it have to be identical to the input signal in terms of amplitude in full feedback? ( I assume that the pot and limiting resistor attenuate the feedback signal when feedback is not at full). Or does it have to be the other way round?

[Kipper4]

As daft and simple as it sounds. Have you tried just wiring the last phaser stage to the first phaser stage with a small series cap? without inverting it?

[Epameinondask]

As daft and simple as it sounds. Have you tried just wiring the last phaser stage to the first phaser stage with a small series cap? without inverting it?

No, but how would this work?

[Mark Hammer]

The suggestion of rolling off the low end is primarily because that's where most of the signal lives.  My own experience is that having more stages also tends to increase the obviousness of the bottom in the feedback.  So, attenuating bass with a shallow 6db/oct rolloff allows the mids and treble of the highly resonant sounds to shine more brightly, without sounding "box-ey".  As well, of there is any small risk for cumulative gain across all stages, the rolloff prevents things from getting out of control.

[Kipper4]

"No, but how would this work?"

Look how Rick did it here (regen)

http://www.diystompboxes.com/smfforum/index.php?topic=80456.0

[Epameinondask]

The suggestion of rolling off the low end is primarily because that's where most of the signal lives.  My own experience is that having more stages also tends to increase the obviousness of the bottom in the feedback.  So, attenuating bass with a shallow 6db/oct rolloff allows the mids and treble of the highly resonant sounds to shine more brightly, without sounding "box-ey".  As well, of there is any small risk for cumulative gain across all stages, the rolloff prevents things from getting out of control.

The smaller cap made the feedback better sounding. But, it's not the sound that I'm after. The last phase stage to second phase stage feedback path that I was using (and like most of the phasers do) is like it just makes the sound more intense in a way that it's almost ready to oscillate.Like an ear piercing wah.  The Small stone also achieves the sound I'm after. It's very thick and rich phaser effect feeding the last phase stage output signal back to the input. So what am I doing wrong here?

[Epameinondask]

"No, but how would this work?"

Look how Rick did it here (regen)

http://www.diystompboxes.com/smfforum/index.php?topic=80456.0

Although I've read this thread and the Causality isn't really the sound I'm after, I'm gonna give it another reading  . Thanks!

[Kipper4]

Sometimes less is more with Regen.
You have to experiment and try to get the regen (feedback) at the level where it sounds more complex than what you had, but no so complex it's a mess.
Have you tried a regen pot wired like an output pot Signal to lug 3 ,Wiper is the output (back to previous stage) and lug1 to ground. with a series cap.

I'm not sure what is best for you.....

[Epameinondask]

Have you tried a regen pot wired like an output pot Signal to lug 3 ,Wiper is the output (back to previous stage) and lug1 to ground. with a series cap.

I'm not sure what is best for you.....

Yes, I've tried that. It's the same effect with wiring the pot as a variable resistor, but a certain point of the pot rotation (towards ground) the phasing effect is gone. It seems that feedback is all about what you're feeding back to where. That's why I'm insisting to make the feedback path as long as possible. It seems that all of the phasers of my taste do the last-phase-stage-to-input feedback path.

I'm doing an LTSPICE analysis on my desing and I see a lot o f interesting things. It seems that the last phase stage output has a wave form slightly out of phase and with waaaaay more amplified compared to the wave form of the point that it is fed back (2nd phase stage inverting input ala Ross Phaser).

Does this have something to say about the feedback signal? Or is there an error in my LTSPICE schematic? Because I do get 3 and 4 notches at the frequency analysis at the output of the phaser. The first notch is deeper than than the others at around 60+db. I've read that deep notches help with the intensity of the phaser and 60db seems fine to me, that's why I'm concentrating on the feedback of the phaser.

I'll do some LTSPICE for the Small Stone to see if there is something that I can find out to help me with the feedback style I'm after.

[Epameinondask]

Well, things kinda start to work out for my design. I just fed the output of the last phase stage back to the input just after the input resistor and I got some feedback. Now here's the new problem 

On the one hand, I can get the kind of feedback that I'm after. Hooray.  Very ehx like. On the other hand, when the feedback trimmer I use (500k with) gets to the point that makes the feedback kick in ,at like 80% of the rotation (I use linear), from that point and until it reaches 100% rotation, it starts to attenuate the overall signal of the circuit. Just like when your guitar volume is very low, almost off.

Does anyone have an idea why is this happening? I can hear the type of feedback that I want but in an attenuated signal. The more I turn the trimmer, the more feedback I get in an even more attenuated signal.

[Epameinondask]

Still haven't been able to tackle this "volume" loss, but here are some interesting things I found out trying to solve this issue. This volume attenuation isn't really a volume loss-attenuation, but in fact it is a mid-treble attenuation.

Here's what's going on. When I use a 500k trimmer and a 1uf capacitor for the feedback path, even at zero ohms of limiting resistance, I get no self oscillation at all. It sounds like the kind of feedback I want, but with almost all the the highs attenuated. It's like all the bass of the guitar signal is there, but the gain of the signal is attenuated, as well as the treble of it. Kinda hard to describe the effect.

When I use smaller cap values and trimmers, this behavior is not that obvious, but it's still there. It still attenuates the gain and frequencies of the signal. This time it introduces self oscillation at some point of the trimmer rotation and the feedback is not that thick and rich. So, lowering cap values doesn't really help.

So, is there is a chance that some sort of a high pass filter is being created with this feedback path? How can I overcome this issue? Would a amplification of the input signal help? (and then maybe attenuate it before it reaches the phase stages?). I'm really running out of ideas 

[Epameinondask]

bump 

[StephenGiles]

I remember trying input frequency dependent voltage controlled feedback to both a phaser and flanger, which I tried back in the 1980s from a suggested circuit I was given, I forget who by, but it used a couple of CA3094s.

The idea was to extract a control voltage proportional to the amount of high frequency content in the input signal, and configured so that with only low frequency in the signal, the amount of feedback would be smallest.

I think it may have been from a guy that serviced my old Teac 3340 reel to reel who lived in Milton Keynes UK, I'll see if I can locate him.

[Epameinondask]

Thanks a lot, Stephen

I remember trying input frequency dependent voltage controlled feedback to both a phaser and flanger, which I tried back in the 1980s from a suggested circuit I was given, I forget who by, but it used a couple of CA3094s.

The idea was to extract a control voltage proportional to the amount of high frequency content in the input signal, and configured so that with only low frequency in the signal, the amount of feedback would be smallest.

I think it may have been from a guy that serviced my old Teac 3340 reel to reel who lived in Milton Keynes UK, I'll see if I can locate him.

I can't say that I fully understand what you are talking about, except from the fact that you tried the same configuration of feedback that I'm trying to use.

What really puzzles me is how this kind of feedback works. For example, the Small Stone phaser (which is a 4 stage phaser), just feeds the signal from the last stage of the phasing stages back to the input, with no fancy way.Just a cap, a limiting resistor and a voltage divider.I tried using only four stages and feedback in the same configuration. Things sounded better, although with my phase stage values the phasing effect sounded more like a wah and at more extreme feedback settings, instead of self oscillation, I got this attenuation that I'm talking about in my previous posts. The more stages I tried using with this kind of feedback, the more intense the gain/frequency attenuation.

What the heck is going on?  . I tried searching for more info in this forum, but all I can find is about feedback with more traditional ways. Like feeding the last phase stage output back to the second stage inverting input.

[StephenGiles]

In simple terms - more high frequency, more feedback, less high frequency less feedback.

[ElectricDruid]

Don't forget that there's positive and negative feedback both possible. Some designs include a switch for this as an option. If the phase-shift stages are based on an inverting configuration, you can swap between them by changing the stage at which the feedback is inserted. If not, you can invert (or not) the actual feedback signal itself.

I also found different results from feeding back the signal from the output mixer (which includes the input) rather than the signal from the last stage (which doesn't). There are a lot of options, and that's *before* you start putting any frequency-dependent tone-shaping components like caps into the audio path.

HTH,
Tom