OJG wrote: My understanding is that a phase shift is really a shift in time. If we have two sine waves, one beginning half a cycle after the other (180 degree phase shift), this would cause them to cancel each other. So all we need to do is move one of these signals in time to avoid this from happening.

Does this make sense ?

Should work fine with two theoretical sine waves, each having just one frequency.

But what about audio signals that are made of many different frequencies ?
Phase cancelation could occur between some frequencies but not others. Then if we time shift one of the signals we may cure phase cancelation in one frequency (and all frequencies that are an integer multiplication of that frequency) but result in phase cancelation between a different group of frequencies.

So what's the solution ?

OJG

I'm not sure how some frquencies could be out of phase and others not short of intential manipulation with crossovers and delays.

phase is a relative measure. Something must be a reference. In electronics it's typically the input signal/spectrum. A non-distorting channel will have a constant group delay (aka linear phase) on its output relative to its input.

Or, one can measure phase shift between frequencies. In loudspeakers/crossovers phase shift between drivers will cause directivity shifts in the main lobes of the speakers pattern (typically a vertical pattern with vertically-aligned drivers).

OJG wrote: ...But what about audio signals that are made of many different frequencies ?
Phase cancelation could occur between some frequencies but not others. Then if we time shift one of the signals we may cure phase cancelation in one frequency (and all frequencies that are an integer multiplication of that frequency) but result in phase cancelation between a different group of frequencies.

So what's the solution ?

OJG

That's fairly close, in that as the combined arrival times diverge from zero, the phase cancellation begins at the highest frequencies and moves down to include lower ones as the time difference increases. And different tones will fall in and out of phase along the way.
Some other factors..
The relative volumes of the two signals determines the depth of the comb filtering. I.e. the 3-1 rule as an aid to hide the phase effects by keeping them lower in volume.
-In the case of a second mic that might be used at near equal volume, you can take advantage of the more random reflections by moving back out of the source's direct sound field.
Or choosing to use the effects as a tonal choice. 8-)
Wayne

phase problems are most problematic when they have least to do with time. most problems have to do when you say; have two mic one one each side of a drum. then your would have positive pressure that would equal a + on the wave form. then on the mic on the opposit side of the drum, you would have negitive pressure that would equal a - on the wave form at the same time as the +. so +1 plus -1 equals 0 and therefore no sound.

frob wrote: you would have negitive pressure that would equal a - on the wave form at the same time as the +. so +1 plus -1 equals 0 and therefore no sound.

good thing they have an instant-fix button for that!

frob wrote: phase problems are most problematic when they have least to do with time. most problems have to do when you say; have two mic one one each side of a drum. then your would have positive pressure that would equal a + on the wave form. then on the mic on the opposit side of the drum, you would have negitive pressure that would equal a - on the wave form at the same time as the +. so +1 plus -1 equals 0 and therefore no sound.

Frob,
In the situation you describe, wouldn't it solve the problem if we time shift (edit) one of those tracks ?
That ought to cancel the phase shift I would think.

OJG

OJG wrote: My understanding is that a phase shift is really a shift in time.

Yes, with respect to some other related waveform. This of course begs the question, "What's the difference then between phase shift and delay?" My opinion is that it depends on the shift's order of magnitude. When the shift is relative to the period of a given waveform, then it is phase shift. When the shift is much greater (>milliseconds) then it is a delay. Phase shifts are perceived as a tonal /spacial coloration, while delays are perceived more as a tempo alteration and/or an echo.

OJG wrote: ...If we have two sine waves, one beginning half a cycle after the other (180 degree phase shift), this would cause them to cancel each other. So all we need to do is move one of these signals in time to avoid this from happening.

Just make sure you understand the difference between a phase shift and polarity inversion. As you mentioned, a phase shift is like a mini delay. But a polarity inversion does NOT introduce any delay. Rather it inverts/reverses the direction of electron flow of the audio signal, thereby reversing the direction of speaker motion. You probably can gather from my explanation that the term "phase" is commonly misused, especially by equipment manufacturers. For example, there is really no such thing as a phase switch on a mic preamp, only a polarity switch.

when trying to think of phase as a delay
please expand your thinking to allow for a
continuously variable delay at different frequencies.

AC theory is not as simple as DC theory and you need to think of all the spectrum, at the same time and not a single frequency
or
as a specific situation ... ( don't just think of the phase sift at 1kHz or at the crossover point etc )
or
as you would with DC theory. ( 0Hz )

A simple perfect resistor has a zero phase shift
but a capacitor and an inductor have a continuously changing phase, opposite to each other and dependant on circuit and load and ... lots of things

put these ideas together and look at a simple speaker driver and things can get very complex.
it amuses me to see people trying to make a passive speaker crossovers from a chart by treating a speaker as an 8 ohm resistor
and then with no consideration to the acoustic output of the system.

Microphone capsules are just as complex.

also when trying to think of phase as a delay remember that altitude and humidity also play a part in this. frankly if you have two mics you ARE going to have some phase abnormalities, but that is not really a bad thing. in fact that is one resone i use more then one mic. i think that you are just over compecating a verry simple thing.

So bottom line --
there will always be some sort of phase shift/cancelation whether we can hear it or not, and unless mixing both signals in mono doesn't cause a major gain reduction, we move on.

OJG

in many respects
it does go with the territory

like all things audio
just deal with problems when they become noticeable

even a simple hi-pass filter has these issues
and
yes even plug-ins will do this
1st , 2nd 3rd order filters all have their respective phase /polarity issues

same goes for passive crossovers ... 6dB, 12db ,18dB etc

shut up Kev !

but a good understanding of the theories is a handy thing
and understanding when to look for a problem ... and when NOT to look for a problem can make life easier.

Kev wrote: but a good understanding of the theories is a handy thing and understanding when to look for a problem ... and when NOT to look for a problem can make life easier.

so true

OJG wrote: [quote=frob]phase problems are most problematic when they have least to do with time. most problems have to do when you say; have two mic one one each side of a drum. then your would have positive pressure that would equal a + on the wave form. then on the mic on the opposit side of the drum, you would have negitive pressure that would equal a - on the wave form at the same time as the +. so +1 plus -1 equals 0 and therefore no sound.

Frob,
In the situation you describe, wouldn't it solve the problem if we time shift (edit) one of those tracks ?
That ought to cancel the phase shift I would think.

OJGI'd expect a time shift would not be the best first fix for a polarity problem, (unless perhaps it was a simple sustained pitch?). Consider the typical uneven waveform- or better yet, a transient. Mostly a large sharp peak. One mic sees it positive, one neg. Where would you shift?

wayne wrote:
I'd expect a time shift would not be the best first fix for a polarity problem, (unless perhaps it was a simple sustained pitch?). Consider the typical uneven waveform- or better yet, a transient. Mostly a large sharp peak. One mic sees it positive, one neg. Where would you shift?

Yes Wayne - I see your point.