How does changing phase by say, 40 degrees work?

[bennychico11]

Geez... this is way overly complicating this. The problem here is, with a single source (a must) how does one go about figuring phase differences with a complex wave like a distorted guitar. The key word is single source. Without it you don't get the frequency enforcement/cancellation due to time (distance) differences... stuff bouncing all over the place and arriving later and/or another mic some distance from the primary. One can really only concentrate on the fundamental frequency and it's harmonics. The rest is irrelevant.

Why would you figure phase differences with a single source? A single wave form cannot be in/out of phase with itself. You're talking about wave interference...which needs two waves. Stuff bouncing all over the place and arriving later is technically TWO waves; the original source and its reflection(s).
I think cpl_crud was alluding to this in saying that yes, each wave form has it's own phase but it's arbitrary and insignificant until you have something else to compare it with. But I think with aperiodic waves, the phase of the wave form is constantly changing, so measuring it would be a huge waste of time and you'd end up with a ton of different numbers for each point in time.

I wouldn't say we're over complicating...I would say we're just talking out the innerworkings of physics, no matter how advanced it may seem it's still the world around us.

 

[Harvey Gerst]

The "phase" problem basically occurs when you're trying to use two mics at different distances from the same source. If both mics are picking up essentially the same signal, there will be a phase difference between the two mics. Running a guitar or bass direct, plus adding a mic can create this phase problem as well. The mic will be out of phase by some degree to the direct signal or in relation to the second mic.

Bill Muellar at PSW added this excellent answer:

"I'm not really sure what you are looking for here. but phase relationships exist between two different points on the SAME WAVEFORM. If you move you mic and overdub another track, it will be a completely different waveform and will have no relationship with the original track other than it will be similar.

However if you are looking for a phase relationship between two microphones on the same amp on the same performance then here is some info.

Sound travels 1128ft/sec at 70 degrees F and sea level. We usually round that to 1130 for calculations. The simple formula is 1130/Distance = frequency. If you want to know the resonant frequency of your room, the formula is 1130/2D=F. That is because a sound must travel from one wall, to the other and back to complete one cycle.

If you want to know which frequency will cancel most perfectly between two points (microphones) then the formula is 565/D=F. That is because sound cancels most perfectly at 180 degrees of the waveform (otherwise known as out of phase).

If you have two mics, one of them exactly 1 foot BEHIND the other relative to the waveform from the guitar cabinet, then they will cancel most perfectly at 565 hz. If one is 6" behind the other they will cancel most perfectly at 1130 hz. If one is 3" behind the other they will cancel most perfectly at 2260 hz. The closer they get, the higher the cancellation frequency becomes until it is out of audio range. Get it?"

 

[Keiffer]

Why would you figure phase differences with a single source? A single wave form cannot be in/out of phase with itself.

I referred to using two mics and to waves arriving late.

 

[bennychico11]

I referred to using two mics and to waves arriving late.

oh, you kept mentioning using only a single source before that

 

[SouthSIDE Glen]

One can really only concentrate on the fundamental frequency and it's harmonics. The rest is irrelevant.

What is the fundamental frequency of a I-IV-V chord riff played on an acoustic guitar? That track is a single complex non-periodic waveform for which - as I now understand it - a phase angle *can* be defined.

Earlier today I would have - and in fact did - agree with you that phase angle is meaningless on such a waveform because there is no fundamental frequency on which to calculate it.

But cpl_crud's dissertations in the other thread that Benny pointed out did indeed remind me (and bring back a flood of old memories) that even non-periodic waveforms - i.e. those without a fundamental frequency - can be plotted in phase space the same way a periodic wave can (they just look messier), and therefore (assumedly) phase angles can equally be calculated.

If that is indeed true, I'm just not sure of the math or fuction one would use to calculate it. In other words, how to answer the question that is the title of this thread for anything other than periodic waves.

It's probably just a Google away, I just haven't had the chance to check it out yet.

G.

 

[kylen]

Just to complicate the complications...

Here's your single-source phase rotator - you can sweep the comb thru the spectrum (and watch it in SPAN if you like - but it is very audible). This also creates the same affect that a standing wave has in your control room so I would suspect the signal is split under the covers and delayed.
http://www.tritonedigital.com/produ...id=33&osCsid=96bae6948560ed9477d730c2c517fa7c

Or you can simply use a stereo delay on a mono signal - same as shifting a cloned track back and forth in time - or using a plugin that has lookahead on one track and not the other (when the host doesn't support delay compensation). Or use a plugin like T-Racks compressor that has a slight phase delay to either channel to give the natural analog sound ( ).
http://www.voxengo.com/product/audiodelay/

But Psycho-Phaser is a little more precise than those guys if you want to sit at one frequency, set the 2 pole all pass, listen to the phase shift create a comb (you know the sound) and watch it in SPAN.
http://braindoc.de/manuals/PsychoPhaserEn.html

Just a coupla different ways to create the natural phenom ITB for fun...

The math and group delay is a little over my head and interest level as a tech- I guess I just worry about if its good or bad, how to create it prevent it in any situation. The sound!

ED: Smelling mistakes...

 

[composition4]

Good to see I'm not the only one who doesn't wholly understand this!

So.. If I have a recording (say, a single snare beat) and I set PhaseBug (which supposedly just delays the signal to achieve phase-shift?) to put it 38 degrees out of phase (compared to the original snare signal)- by how many milliseconds/samples would it delay it? How would the program quantify how much to delay it by??

Sorry if it seems like I'm going around in circles here, but I'm in the same boat as SouthSideGlen - I can't see how you could determine the wavelength of such a complex waveform. And you need wavelength to calculate phase-shift, right?

Thanks guys,
Jon

p.s. I just gave up yesterday, my brain couldn't take it any more.

 

[bennychico11]

by how many milliseconds/samples would it delay it? How would the program quantify how much to delay it by??

that's the thing...we don't know. How the programming actually works is a mystery to us. It could be calling itself a phase shift when it's actually just delaying it by a certain degree....or it could be referencing a standard frequency that we don't know about. The programmers know how it works.

I guess it comes down to....ignore the theory behind it and ask yourself does it sound good?
We just took this whole thread and modulated it into "how does angular phase shift look in the complex plane"

 

[FALKEN]

you don't need to know the wave-length in order to talk about phase.

when talking about voltage, there is no such thing as absolute phase. we can only talk about phase in a relative sense; there must be a reference waveform, usually the power supply waveform.

for a reference, consider a perfect sine wave. it starts at 0 degrees, peaks at 90 degrees, comes back down to the horizontal plane at 180 degrees, has a negative peak at 270 degrees, and returns to the starting point at 360 degrees.

if a wave were "at 40 degrees phase" it simply means that it begins at a point between the horizontal and the peak (somewhere between a sine wave and a cosine wave). If a wave were shifted by 40 degrees, this is not a change in time!! it is a re-drawing of the wave so that it starts above the horizontal and ends above the horizontal.

this does not change the properties of the wave. it is still the same frequency and amplitude, and still exists in the same time-space. it just begins and ends at 40 degrees (relative to the reference) instead of 0 degrees.

I hope this makes sense.

 

[SouthSIDE Glen]

this does not change the properties of the wave. it is still the same frequency and amplitude, and still exists in the same time-space. it just begins and ends at 40 degrees (relative to the reference) instead of 0 degrees.

I hope this makes sense.

Yeah, that makes total sense, and is something I think we all agree upon as a basis. But the question remains as to just how does one calculate the phase when there is no reference. That's where we're at with this.

Your description of the classic angles in a sine wave implicitly identifies "degree" as 1/360th of one full sine wave; that is, it is indeed related to the wavelength when looked at that way.

That approach, however, falls short when it comes to calculating amount of phase angle on a complex, non-periodic wave. Without being able to say that one degree is 1/360th of something - on a non-periodic wave there is no "something" on which to base the identity of one degree.

Yet such real life waveforms can be graphed in phase space just like a sine wave can. Therefore they can also be rotated through phase the same way. You could take a recording of my voice singing this post and rotate it 40° in phase (just rotate the plot in phase space by 40 degrees.) even though my voice has no "reference" frequency or wavelength.

The question remains as to how one does that. I was hoping someone else may have picked up the ball on that one by now, but it hasn't happened. I'll have to reasearch that during lunch today.

G.

 

[composition4]

Thanks again for the replies. Glen, be sure to let us all know when you learn a bit of information.. I'll see what I can find out also.

Jon

 

[kylen]

...So.. If I have a recording (say, a single snare beat) and I set PhaseBug (which supposedly just delays the signal to achieve phase-shift?) to put it 38 degrees out of phase (compared to the original snare signal)- by how many milliseconds/samples would it delay it? How would the program quantify how much to delay it by??
...

Here's another take on all of this. I don't do much theorizing out here in the hollar so I'm coming in from some tools I play with and other forums where folks have discussed things of this nature.

A couple of things that have been mentioned, more or less (ref Rane):
phase angle indicates a value in time relative to an arbitrary crossover point (usually 0 degrees) thru which the periodic input waveform has progressed.

When two signals (ie input and output) have the same exact time relationship to each other, they are said to be in-phase

Changing the phase angle also changes the time value at any given frequency.

A phase-shifted sine wave appears displaced in time from the input waveform. This displacement is called phase delay (or group delay) and is [usually] constant for all frequencies of interest.

Voxengo makes a tool that I think was modeled from a hardware box, however the Voxengo plugin is phase-linear:
The PHA-979 allows "arbitrary phase shift to the program material. What is meant by the phase shift here is a simultaneous shifting of all frequencies across the active frequency range of the signal by the given phase shift value. This process does not harm the phase relations within the signal being processed."

Ih his forum at Voxengo Aleksey goes on to explain how phase angle relates to time in this tool:
( http://www.voxengo.com/forum/pha979/849/ )
..."...For example, if you have a 64 Hz tone, its full period lasts 15 ms. So, 180 degree shift translates to 7.5 ms shift in time. 90 degree shift is the same as 3.75 ms shift. Phase shifting PHA-979 performs is pretty the same as splitting the signal to infinite number of bands and then time shifting each band for the amount depending on the band's frequency."

"The most obvious example of how phase shifting differs from time shifting: for example, if you want to time-align bass you will need to shift the track pretty far (several milliseconds). For the same amount you will shift the higher frequencies. By using phase-shifting you may shift lower frequencies in time for the same amount (several milliseconds), while higher frequencies won't be shifted in time by any considerable amount. Again, PHA-979 is *not* a delay unit. You may check this out yourself."

"By the way, if you understand the purpose of the phase invert switch available on mixers and mics, you may understand the purpose of PHA-979. PHA-979 simply offers 178 intermediate positions to this 'phase invert' switch."


Anyway - I think using the PHA-979 tool and comparing it to results with some kind of track delay (I mentioned earlier) the difference should be obvious in sound. I'll have to check this out myself. As far as what a shift in phase angle has to do with time at any given frequency I guess you would have to calculate it since - even though it may be constant - time value is also based on frequency (obviously).

We don't normally think in terms of "I think I'll try a time delay of 0.02 ms at 2KHz to fix this" do we? What are we doing here again?

 

[SouthSIDE Glen]

Phase shifting PHA-979 performs is pretty the same as splitting the signal to infinite number of bands and then time shifting each band for the amount depending on the band's frequency."

"The most obvious example of how phase shifting differs from time shifting: for example, if you want to time-align bass you will need to shift the track pretty far (several milliseconds). For the same amount you will shift the higher frequencies. By using phase-shifting you may shift lower frequencies in time for the same amount (several milliseconds), while higher frequencies won't be shifted in time by any considerable amount. Again, PHA-979 is *not* a delay unit. You may check this out yourself."

Ah HA! OK, now I get it...I think . And if I do, that also describes the detail difference between a phase shift and a time shift.

Try this out and see if it makes any sense or if I'm still way off...

If I read that right, what it's saying is that it's phase shifter is actually breaking down the waveform into a continuum of component frequencies or frequency bands (probably via a Fourier series), much like a spectrum analyzer does. But instead of just displaying amplitudes on a graph like a spectrum analyzer, it's actually then applying a phase shift to each component frequency band by x number of degrees, with the size/definition of the degree being defined by the particular component frequency which it's currently processing.

In other words, there *is* a reference frequency used to define the value for "degree" of phase shift, the catch is that that reference frequency changes to reflect the frequency band on which the phase shift is being applied.

A 40° phase shift on a complex wave form means that there is a 40° phase shift at 20Hz based upon the wavelenth at 20Hz, and also a 40° phase shift at 20kHz, but there the value for one degree of shift is based upon the wavelength at a frequency of 20kHz, not 20Hz. Repeat as necessary for all frequencies in between.

That differs from a standard time shift in that the time shift is applied to the entirety of the original complex waveform, regardless of component frequency in which case "degree" is indeed meaningless. In other words, such a shift is just a simple delay (of course!)

I hope I have that right, because that falls into place perfectly for me and seems to make total sense. It also means that we all were right all along, really, we were just missing the big picture.

Does that make as much sense to every one else as it does to me? I hope so, because my brain was staring to hurt . And thanks a bunch for digging that up kylen!

G.

 

[kylen]

...A 40° phase shift on a complex wave form means that there is a 40° phase shift at 20Hz based upon a wavelenth of 20Hz, and also a 40° phase shift at 20kHz, but there the value for one degree of shift is based upon a fequency of 20kHz, not 20Hz. Repeat as necessary for all frequencies in between...

That's my take on all of this too Glen...applying a straight 40 degrees phase shift (group delay) is the same shift in phase degrees at 20Hz as it is at 20KHz, however the time difference is not the same since we have different periods to plug into the formula. Kind of like an octave at 100Hz and an octave at 10KHz are still an octave - but there's a lot more bandwith (Hz) in the one at 10KHz...

What's heavier a ton of pillows or a ton of bricks...I guess we could go there next!

 

[SouthSIDE Glen]

What's heavier a ton of pillows or a ton of bricks...I guess we could go there next!

You take the bricks, I'll take the pillows. Much better for resting my head upon after racking my poor tired brain over this question for the past two days

I feel like you've lifted a ton of brick pillows off my head. Thanks again .

G.

 

[kylen]

Just returning the favor...whenever possible - not often enough!

 

[FALKEN]

I think you guys are over-thinking it. if you are shifting the phase by 40 degrees then the reference would be the original wave. it doesn't matter if it is a complex wave or a simple one.

 

[SouthSIDE Glen]

I think you guys are over-thinking it. if you are shifting the phase by 40 degrees then the reference would be the original wave. it doesn't matter if it is a complex wave or a simple one.

It does because one needs do define just how much of a shift 40° actually is. In order to do that, one needs to define just how far of a shift one degree would be; they need to define a degree. The amount of one degree is 1/360th of a wavelenth. What's the wavelength of a complex wave?

It has no fundamental wavelength, and therefore no fundamental measurement on which to base the amount of one "degree" of phase. The wave needs to be broken down into it's constituant frequency sine waves via Fourier transform (the same way that a spectrum analyzer breaks a complex waveform down to it's constituant frequencies.) Then one can define "degree" as being 1/360th of the wavelength of each frequency band component of the complex waveform. The key here is that the amout that one degree shifts with the component frequency; because wavelength shortens as frequency increases, the actual amount of physical delay for the high frequencies is actually shorter than it is for the low frequencies, even though the 40° measurement remains constant. This is not just a simple pushing of the waveform down the timeline.

For a pure frequency sine wave the result is the same whether you do the Fourior transform or just the simple shift, it's OK to ignore the transform in one's thinking because the fundamental measurement of degree is static, easily defined by the sine wave's wavelength. But when you move to complex waves it makes a big difference because there is no simple wavelenth on which to caclulate.

G.

 

[bennychico11]

It does because one needs do define just how much of a shift 40° actually is. In order to do that, one needs to define just how far of a shift one degree would be; they need to define a degree. The amount of one degree is 1/360th of a wavelenth. What's the wavelength of a complex wave?

yep, I used to think that phase=time. But after this discussion and the one I linked to back on the other page, I think it's the other way around...time is a component of phase. And there are many other components at work to define the phase shift of a waveform.

I'm glad you guys are working this out because it's interesting. And come to find out, this is pretty much exactly how I explained the phaser effect works. With a phaser:

The degree of phase shift is modulated periodically....so you really can't pinpoint and say the two waves are "40 degrees out of phase," because it varies continuously.

This is between two waves, one of them that has stayed the same and a copy that has been modulated....but it's pretty similar to a typical phase shift of a single, aperiodic wave.