MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

Phase mod vs. phase shift?
MUFF WIGGLER Forum Index -> Eurorack Modules Goto page 1, 2, 3  Next [all]
Author Phase mod vs. phase shift?
Devilwidget
I am trying to find a module that will phase modulate an input. There are a range of phase shifters, but none of them seem to work as I desire - presumably phase shift is a different process? I want something like the Richter Osc 2, or Livewire AFG (i.e. move the waveform's phase without changing its shape). All of the 'phaser' modules seem to change the shape of the waveform, not its phase position. What module can do this for external signals (i.e. not just to the waveform of its own vco)?

Thanks!
mbartkow
So you basically need a voltage controlled delay that is capable of very short delays and that allows to control the delay very precisely, down to a fraction of the cycle of your input waveform. Like the d0 from Mungo, perhaps

Explanation: delay is a form of linear phase shift. For a complex signal the amount of shift applied to its spectral components is linearly dependent on their frequencies. Only this type of phase shift offers waveform preservation. Any other (i.e. non-linear) phase shifting device will introduce distortion due to the inevitable dispersion
BaloErets
WMD made this just for you thumbs up
starthief
Devilwidget wrote:
All of the 'phaser' modules seem to change the shape of the waveform, not its phase position.


A "phaser" is an effect based on changing the phase via a delay -- and then modulating it, feeding it back for more shifting, and mixing it with the original signal.

Between the emphasis from feedback and the phase cancellation from mixing it with the original, that's why the shape changes. There's also sort of a "detuning" effect, because modulating the phase changes the frequency (it's pushing peaks and troughs closer together, or spreading them out).

Usually for phase modulation, if I'm not doing it on the E370 or Hertz Donut with their internal features, I use Disting's delay algorithm. It just requires heavily attenuated CV, a short delay time and no feedback.
Dogma
i think of it simply like this. Its not exactly right but right enough


Phase = time
Frequency = pitch

Right now in LFO mode im phase modulating the delay of my modcan DD and using the phase modulated outputs vs the non phase modulated (ie top riw vs bottom row on the richter) theres a world of sounds. Modulate the phase mod input with its own waveform even more.....

Im a phase nut - i have the PDO, Cwejman PH8, Richter, ASSIMILTOR, modcan quad, omiga phi II - all phase modulateable....
Graham Hinton
starthief wrote:

A "phaser" is an effect based on changing the phase via a delay -- and then modulating it, feeding it back for more shifting, and mixing it with the original signal.


That is a "flanger", originally created by using tape delay and modulating the speed by pressing the reel flanges. A phaser is a multi-stage all-pass filter mixed with the original signal and produces less nulls.

Quote:

Between the emphasis from feedback and the phase cancellation from mixing it with the original, that's why the shape changes.


The wave shape changes with a phaser because not all harmonics are affected equally.
vidret
Since a ‘phaser’ would be the opposite of a band pass filter (a notch filter) you could prob get there by taking your signal through a bp filter, invert it (maths? Maths.) and then mix the inverted signal with your original.

Not tested, but in theory this should get you there.
starthief
d'oh! Yeah, I do keep mixing up "phaser" and "flanger".

(From my shallow wading into DSP, I'm used to thinking of filters as delays, and everything happens as a result of phase cancellation.)
pugix
BaloErets wrote:
WMD made this just for you thumbs up


The PDO will do it. I don't know of a phase offset module. Some VCOs have quadrature sine outputs, separated by 90 degrees. The PDO allows control of the phase separation. The Mankato VCF has eight different sine wave phase outputs.

As for phase displacing complex waveforms, I can't say that I understand it. Presumably the PDO does this.

A simple inverting mixer will get you a 180 degree 'shift'. smile
mbartkow
People often confuse phase shifting with the function of a phaser effect. A phaser combines a phase shifted signal with original one to get notches is frequency response due to cancellation. I believe the OP didn't ask for any cancellation, the request was very clear for shifting a whole waveform in phase without distorting the waveshape.

PDO is an oscillator, it does not phase shift an arbitrary input signal, and the OP asked for this.

The only way to get a phase shift without distorting the wave shape is to shift every spectral component by the amount (expressed in degrees or radians) that is linearly proportional to the frequency, there is no other way. And such a shift is nothing else than a pure delay - every part of the signal is delayed by the same time (expressed in seconds).

The relationship between phase and time is very easy to understand, once you get this simple example: If you delay a 1Hz sinusoid by 1/8s you get a phase shift of 45degs, but for a 2Hz sinusoid the same delay will give you 90degs of phase shift.

Btw, inverting the sign is not equivalent to phase shift by 180 degrees, it is a common fallacy. This works only for waveforms that are half-wave symmetric, like a sinusoid. For example, a ramp shifted by half of its period (which is the true 180degs) is not the same as ramp inverted.
Devilwidget
This thread has been enlightening, thanks all!

Question then - does a delay not implicitly degrade the source? Would it have to be a digital delay to phase modulate (or phase shift, which seems more accurate, despite the realty of phase shift modules), without changing the waveform?
Dcramer
Devilwidget wrote:
This thread has been enlightening, thanks all!

Question then - does a delay not implicitly degrade the source? Would it have to be a digital delay to phase modulate (or phase shift, which seems more accurate, despite the realty of phase shift modules), without changing the waveform?


Yup, I think so, but then again I’m not too bright.

What I’m wondering is what do you want this for?
If you slowly modulated an external oscillators phase with a short, clean, delay, wouldn’t you just get a little bit of vibrato? hmmm.....
authorless
If you want phase shifted waveforms, you are pretty much stuck with an oscillator that does it natively. OP mentioned the Oscillator II (which does) and the AFG (which does not).
tbecker
mbartkow wrote:
People often confuse phase shifting with the function of a phaser effect. A phaser combines a phase shifted signal with original one to get notches is frequency response due to cancellation. I believe the OP didn't ask for any cancellation, the request was very clear for shifting a whole waveform in phase without distorting the waveshape.

PDO is an oscillator, it does not phase shift an arbitrary input signal, and the OP asked for this.

The only way to get a phase shift without distorting the wave shape is to shift every spectral component by the amount (expressed in degrees or radians) that is linearly proportional to the frequency, there is no other way. And such a shift is nothing else than a pure delay - every part of the signal is delayed by the same time (expressed in seconds).

The relationship between phase and time is very easy to understand, once you get this simple example: If you delay a 1Hz sinusoid by 1/8s you get a phase shift of 45degs, but for a 2Hz sinusoid the same delay will give you 90degs of phase shift.

Btw, inverting the sign is not equivalent to phase shift by 180 degrees, it is a common fallacy. This works only for waveforms that are half-wave symmetric, like a sinusoid. For example, a ramp shifted by half of its period (which is the true 180degs) is not the same as ramp inverted.


This all assums that the signal is periodic, if it isn't than there is no phase with respect to the signal! Some really cool sounds take this form with asymetric transients and usually sound drum-like.

The allpass filters that were mentioned offer phase shift with repect to the frequency range of the phaser, meaning that for a nonperiodic sound the phaser can only shift the periodic parts of the spectrum. The classic effect is this allpass setup with some mixing, feedback and a variable lfo that governs the filter center.
mbartkow
tbecker wrote:

This all assums that the signal is periodic, if it isn't than there is no phase with respect to the signal!

That's not entirely true. The existence of continuous Fourier transform (which is invertible) proves that every (even non-periodic) signal consists of sinusoids. For transient signals the spectrum does not consist of individual harmonics, but is a continuous complex-valued function X(w), usually decaying to zero with frequency. Also, a so called phase spectrum can be computed as the arg of X(w). Any delay-related signal processing has an impact on the phase spectrum that can be precisely computed. A phase relationship between two non-periodic signals can be computed as well as the difference between the corresponding phase spectra. If the signals do not have the same shape, this phase difference will be a non-linear function of frequency, that's all.

tbecker wrote:

The allpass filters that were mentioned offer phase shift with repect to the frequency range of the phaser, meaning that for a nonperiodic sound the phaser can only shift the periodic parts of the spectrum.

I'm sorry, but this is not true at all. Allpass filters affect all signal components, periodic and non-periodic. You can use an oscilloscope to learn that. There is a so called "group delay" which is the derivative of phase response with respect to frequency. The problem with allpass filters is that their phase response is a nonlinear function of frequency, hence the derivative is not constant. And that means the filter is dispersive, it delays various frequencies with different amount, which yields waveform shape distortion.
tbecker
[quote="mbartkow"]
tbecker wrote:

This all assums that the signal is periodic, if it isn't than there is no phase with respect to the signal!

That's not entirely true. The existence of continuous Fourier transform (which is invertible) proves that every (even non-periodic) signal consists of sinusoids. For transient signals the spectrum does not consist of individual harmonics, but is a continuous complex-valued function X(w), usually decaying to zero with frequency. Also, a so called phase spectrum can be computed as the arg of X(w). Any delay-related signal processing has an impact on the phase spectrum that can be precisely computed. A phase relationship between two non-periodic signals can be computed as well as the difference between the corresponding phase spectra. If the signals do not have the same shape, this phase difference will be a non-linear function of frequency, that's all.

[quote="tbecker"]

The continuous Fourier transform theoretically proves for every signal it can be constructed by sinusoids, however this is done with greater complexity for certain nonperiodic sounds and in practice is approximated for these same nonperiodic waveforms in dsp. If the continouos Wavlet transform were used instead for nonperiodic waveforms the respresentation make more sense, but then the idea of a periodic phase in the Fourier is given up for the Wavlets position or time components.

Because people don't actually hear absolute phase at all, we are really talking about phase difference and since nonperiodic sounds have specific components in specific positions and the concept of phase doesn't fit, these sounds are simply moved in time position for the phase effect (as you have mentioned by using a variable delay line).

Anyway thanks for your input on allpass filters, I assumed they should be linear to be usefull in the context of phase change.
Devilwidget
My initial reasoning for asking the question was because I wanted to explore phase in more detail, and it was confusing me that 'phase shift' and 'phase mod' and 'all-pass filter' seemed to be used changeably by manufacturers. There have been so many in depth and helpful replies that I am extremely glad I asked, lots of food for thought here!
Rex Coil 7
For experimenting and learning, I would imagine going with a VST (or perhaps more than one) would be my first instinct. I have no idea if there are such VSTs offered, but I'd certainly begin my search into this topic there.

Far more cost effective if nothing else.

I know that using software wouldn't be as sexy or cool as a hardware solution, however if the whole thing turns out to be a bust, you'll have less money tied up in the project.

I am waiting on the delivery of a Nord Micro Modular, this will be the 5th one I've owned over the decades (I'm keeping this one, by god!). So obviously that thing is forward most in my mind at this time. That said, I have to wonder if the Nord Modular (any one of them) is capable of doing such a thing? And if not, you would still own one of the most badass VA synths ever made. Certainly a hyper powerful learning/experimenting tool if nothing else.

You might also give a look/see at the Nord "G2" demo software. It's 100% free (no money needed), and offers a shit ton of capability ... no support hardware needed other than a half assed capable computer. The demo version is monophonic only, and a few of the modules that are available in the hardware version are not accessible in the demo, but still, the Nord G2 demo software may just be able to provide you with the tools required to stick your toe in the water.

seriously, i just don't get it Just a suggestion.
ibzieg
Rex Coil 7 wrote:

I am waiting on the delivery of a Nord Micro Modular, this will be the 5th one I've owned over the decades (I'm keeping this one, by god!). So obviously that thing is forward most in my mind at this time. That said, I have to wonder if the Nord Modular (any one of them) is capable of doing such a thing? And if not, you would still own one of the most badass VA synths ever made. Certainly a hyper powerful learning/experimenting tool if nothing else.


The G2 has a simple fixed-time delay module you can use shift the phase of a signal backwards. I think you can even specify the time in samples if needed. There is also a simple phase-mod oscillator.

Not sure if these are available on the Micro Modular, mine has a broken power jack and I haven't used it for a while...
Devilwidget
thanks for the software advice, but I spend most of the day looking at a screen - eurorack is my escape!
Rex Coil 7
ibzieg wrote:
Rex Coil 7 wrote:

I am waiting on the delivery of a Nord Micro Modular, this will be the 5th one I've owned over the decades (I'm keeping this one, by god!). So obviously that thing is forward most in my mind at this time. That said, I have to wonder if the Nord Modular (any one of them) is capable of doing such a thing? And if not, you would still own one of the most badass VA synths ever made. Certainly a hyper powerful learning/experimenting tool if nothing else.


The G2 has a simple fixed-time delay module you can use shift the phase of a signal backwards. I think you can even specify the time in samples if needed. There is also a simple phase-mod oscillator.

Not sure if these are available on the Micro Modular, mine has a broken power jack and I haven't used it for a while...
Damn ... these Nords ... what a peach! So much power.

Devilwidget wrote:
thanks for the software advice, but I spend most of the day looking at a screen - eurorack is my escape!
Gotchya .. I just didn't want you to spend money on something that may turn out to be a bust. With software, you can at least TRY the notion out ... and THEN spend a few dollars on the correct hardware should software trials work out.

But I understand your wishes. Carry on.

cool
Navs
Devilwidget wrote:
... I want something like the Richter Osc 2, or Livewire AFG (i.e. move the waveform's phase without changing its shape).


The Doepfer A-137-2 Sawtooth Animator will do this. The effect it's built for is the 'supersaw' sound, but you can remove phase shifted stages from the mix. It will output saws or squares and you can use it at audio or LFO/clock rates for funky rhythms:

https://navsmodularlab.blogspot.com/search?q=a-137-2

edit: I have one for sale, if you're interested wink
tbecker
Navs wrote:
Devilwidget wrote:
... I want something like the Richter Osc 2, or Livewire AFG (i.e. move the waveform's phase without changing its shape).


The Doepfer A-137-2 Sawtooth Animator will do this. The effect it's built for is the 'supersaw' sound, but you can remove phase shifted stages from the mix. It will output saws or squares and you can use it at audio or LFO/clock rates for funky rhythms:

https://navsmodularlab.blogspot.com/search?q=a-137-2

edit: I have one for sale, if you're interested wink


Don't these usually require certain shapes, such as saw?
flashheart
That's what I thought, certainly saw, not sure how it would work with a triangle.
Navs
Yes, they need a saw as input. If you want a phase-shifted triangle or sine, you'd need to follow the output with a waveshaper.

Devilwidget, what is it that you want to do?
MUFF WIGGLER Forum Index -> Eurorack Modules Goto page 1, 2, 3  Next [all]
Page 1 of 3
Powered by phpBB © phpBB Group