kubeek
New member
Sorry, wrong numbers off course.. Should have been -10/10.
SouthSIDE Glen
independentrecording.net
Noisewreck is right, DC offset and waveform asymmetry are NOT THE SAME THING. Nor am I saying the are. Those sax waveforms noisewreck is showing are asynnetrical but they have no DC offset. the "rest point" at the beginning and end is still 0DC.
What I am saying is that if you take those waveforms and invert their phase (not their polarity), that the entire waveform will be flipped, but will stay in place on the Y axis; i.e. it will be upside down, but will not move overall in the vertical scale.
But since the waveform is asymmetrical, the "centerline" around which that wavefrom will flip is not the same as the actual centerline of the chart (ithe wave is not centered around 0DC.) The offset, if you will, is in the location of the mirror itself.
Therefore, the rest line of that waveform waveforms WILL move when the phase alone is changed, and an offset will therefore be created by the phase flip.
Flip the polarity, however, and the offset is not created because the flipping is around that 0DC rest line.
All I've been trying to say all along is that because of this offset in the location of the "mirror" depending upon whether one is changing phase by flipping polarity or whether one is changing phase irrespective of polarity, that the results are different; that they are two different physical processes.
Upon reflection (bad pun extremely intended
), there may not be anythng more than an incedental difference in sound because of physical loudspeaker limitations like have been mentioned earlier. But let's assume there is no theoretical difference in sound whatsoever:
That doesn't change my basic argument that if one wants to understand what a phase switch actually *does*, they need to understand that it is flipping polarity, not just flipping phase like the nomenclature suggests.
THAT'S ALL I HAVE BEEN REALLY TRYING TO EXPLAIN!
---
Here's further consideration that I'm admittedly a little shakey about, but it may add another dimension to the importance of the difference: While the wave itself may not sound different, the difference in average voltage levels will affect what can be done with the mix in which that track is used.
For an extreme example, imageine a 1kHz pulse wave with 10% pulse width modulation that ranges from 0DC at rest to +10dB on top of the pulses. While the signal decreases headroom by 10dB, actually 90% of the time the signal is at 0DC.
Now invert the phase. Now the pulse is inverted, but the voltage range remains the same. Now the signal is "resting" at +10dB and the pulses are "peaking" downwards to 0DC.
Let's assume we have the perfect loudspeaker that won't distort because of the new 10dB DC bias; the signal will probably sound the same (golden ears will argue against that, but let's ignore that angle for now). *However", there is an important difference if we use this phase inverted but not polarity inverted track in our mix. While the overall headroom between the two states remains the same, now that headroom is being taken away not 10% of the time, but 90% of the time. It kinda simulates a denser track (in a very loose way); the RMS does not change like it would in a denser track, but because of the new offset the amount of "space" is reduced.
G.
What I am saying is that if you take those waveforms and invert their phase (not their polarity), that the entire waveform will be flipped, but will stay in place on the Y axis; i.e. it will be upside down, but will not move overall in the vertical scale.
But since the waveform is asymmetrical, the "centerline" around which that wavefrom will flip is not the same as the actual centerline of the chart (ithe wave is not centered around 0DC.) The offset, if you will, is in the location of the mirror itself.
Therefore, the rest line of that waveform waveforms WILL move when the phase alone is changed, and an offset will therefore be created by the phase flip.
Flip the polarity, however, and the offset is not created because the flipping is around that 0DC rest line.
All I've been trying to say all along is that because of this offset in the location of the "mirror" depending upon whether one is changing phase by flipping polarity or whether one is changing phase irrespective of polarity, that the results are different; that they are two different physical processes.
Upon reflection (bad pun extremely intended
), there may not be anythng more than an incedental difference in sound because of physical loudspeaker limitations like have been mentioned earlier. But let's assume there is no theoretical difference in sound whatsoever:That doesn't change my basic argument that if one wants to understand what a phase switch actually *does*, they need to understand that it is flipping polarity, not just flipping phase like the nomenclature suggests.
THAT'S ALL I HAVE BEEN REALLY TRYING TO EXPLAIN!
---
Here's further consideration that I'm admittedly a little shakey about, but it may add another dimension to the importance of the difference: While the wave itself may not sound different, the difference in average voltage levels will affect what can be done with the mix in which that track is used.
For an extreme example, imageine a 1kHz pulse wave with 10% pulse width modulation that ranges from 0DC at rest to +10dB on top of the pulses. While the signal decreases headroom by 10dB, actually 90% of the time the signal is at 0DC.
Now invert the phase. Now the pulse is inverted, but the voltage range remains the same. Now the signal is "resting" at +10dB and the pulses are "peaking" downwards to 0DC.
Let's assume we have the perfect loudspeaker that won't distort because of the new 10dB DC bias; the signal will probably sound the same (golden ears will argue against that, but let's ignore that angle for now). *However", there is an important difference if we use this phase inverted but not polarity inverted track in our mix. While the overall headroom between the two states remains the same, now that headroom is being taken away not 10% of the time, but 90% of the time. It kinda simulates a denser track (in a very loose way); the RMS does not change like it would in a denser track, but because of the new offset the amount of "space" is reduced.
G.
Attachments
Last edited:
noisewreck
New member
Nope. First of all, you never record a Sax by sticking a mic into the bell. And if it was properly recorded, then there would be no such phenomenon. I can produce similarly asymmetrical waveforms electronically, just by using filtering and EQ. The waveform is an inherent propery of the formants of the Sax.
noisewreck
New member
Man, I can't wait to get home to try this experiment.
SSG, from your last screenshot then your premise is that if one phase reverses an asymmetrical waveform, the inversion will introduce a DC offset, while polarity inversion will not.
Interesting. I am pretty sure that Cubase's "Phase reverse" will NOT introduce DC offset.
SSG, from your last screenshot then your premise is that if one phase reverses an asymmetrical waveform, the inversion will introduce a DC offset, while polarity inversion will not.
Interesting. I am pretty sure that Cubase's "Phase reverse" will NOT introduce DC offset.
SouthSIDE Glen
independentrecording.net
That's how I understand it. The CORE of that, and what I've been saying from page one is that because phase change only results in such nonsequitors, that phase change alone is not really what the "phase invert" buttons on our gear actually do. What they actually do is invert polarity, which is a different animal that inverts around a different axis. And that's the part that - while I have plenty of people disagreeing with my assertation, so far absolutely nobody has come forward with a description or an explanation of where or why I'm wrong in saying that.
TAKING IT BACK HOME
To take it all the way back to the first page, and even moreso the earlier thread on this topic, phase is nothing but a measurement of change in phase angle of an original waveform. Whether it's 30° or 300° doesn't matter, it's just a measurement of rotation through a cycle of 360°. The fact that a change of 180° happens to superficially resemble a polarity inversion doesn't make a phase change of 180° anything "special" when looked at in the light of phase alone. It's just a different measurement of rotation. Whther it's time dependent or not (i.e. a shift or an inversion) is irrelevant, it's still just a measurement of a 180° difference in relation to the original wave. There is no change in the voltage envelope no matter what that measurement is.
There is also no relation to any particular arbitrary reference voltage. Phase change retains the same properties and intrinsic boundaries whether it happens near 0DC or -250,000dB.
That's what its being illustrated in the third part of that illustration; a phase inversion alone is nothing more than a phase rotation, or shift, of 180°, except that it is not shifting left or right in time on the graph, rather it is a hypotheticaly instantaneous change in phase value that cares not about the 0DC line.
AND BRINGING IT BACK HERE AGAIN
We all know that the offset shown in "C." does not really happen in real life when we hit the "invert" button; 0DC remains the rest mode (silence) no matter what. That's because we are not execuiting a phase change per se, we are instead executing an inversion of polarity.
In additional fact - at least as far as I understand it - instantaneous phase inversion of a complex asymmetrical wave is only possible in a non-linear environment like a digital editor, where the entire waveform envelope can be examined. But an on-the-fly phase change only, as implied by the label on an analog "phase invert" button is actually just not possible. There is no look-ahead ability in a time-dependant analog situation like an analog mixer or a microphone, therefore no ability to either determine the true offset of the resulting waveform, and therefore therefore no ability to invert the phase in place.
Instead what the microphone button does - and I assume the mixer, and I assume any software emulations of those inverter buttons do - is simply invert the polarity of the signal, just like they do for running a balanced line. It's a simple change of sineage on the values. And that's what we see in B.
I'll bet you that you are exactly right about that. Not because phase change and polarity change are the same thing, because we can plainly see that they are not.
Instead, it does not cause the offset because it's not really executing a phase change, it's executing a polarity change.
Which is why, by technical definition, "phase inverter" is a misnomer for that button. Which is the only point I've been trying to make this whooooole stupid time
.G.
noisewreck
New member
Well, I said the same thing on the first page too. So, who's disagreeing with us? Let's pick up a couple of sticks and stick it to them... uh... beat them up... uh... I need more drinks.
J
jrhager84
expert newb
Sha fuckin' poopie!!!
That is all gentlemen. I thought I'd contribute something to this thread, so there ya have it...
That is all gentlemen. I thought I'd contribute something to this thread, so there ya have it...
SouthSIDE Glen
independentrecording.net
Well, I'm still waiting for RobD to drop the other shoe, and he's got the backing of Benny in this thread so far.
But the history of this topic on this forum is much wider than that. In general, there seems to be a general "understanding" that phase and polarity inversion are interchangable concepts. They aren't. And the instant anybody says otherwise - which is extremely rare because it's perceived as a "the emperor wears no clothes" situation which draws the ire of the crowd when it's brought up - it winds up causing 9 page threads.
The problem is that I have yet to see in those 9 page threads - even after repeated requests by many - any actual positive explanation for why the polarity/phase equalists think they're right, other than the faith that the label on the button is technically accurate, and that dealing with regular AC current, they are indeed basically the same, so it must apply to auio processing as well.
It's just a bunch of circular argumentation that basically says, "when I push the phase button it performs a polarity inversion, so phase and polarity must be the same thing. And to say otherwise is wrong because if it were right, the button would be labeled differently."
---
To jrhager:
Your phase questions were answered back on the first page, what are you still missing? You said you couldn't continue with the phase issue anyway until you get the clicks thing worked out, which is an entirely different question and situation. I haven't seen that thread anywhere. Did you get that worked out yet? Have you considered asking that question in a thread that isn't about a "Question about phase reversal"?
You might want to start by making sure your sample rates are set the same everywhere through the digital part of your signal chain from your converters through your plugs, that all tracks are recorded at the same sample rate, that you're not dropping any samples when recording to your hard drive because of system performance issues (get rid of all your TSRs, for example). Take that info, run with it, and if you still don't have an answer to your clicking, open a thread on that topic and ask about it. But that most likely has nothing to do with any questions about phase reversal.
Oh, and one more recommendation: remove half the mics you're using to record your drums and you'l get rid of most of your phase issues altogether. If your drummer is halfway decent and sounds fine to your two ears when you stand in front of the stage when he's playing, there's no good reason why you can't get an equally good sound to stick to disc with only 3 or 4 mics total. If you can't get a decent sound from stereo OHs and a kick mic (with a snare mic optionally but not necessarily included), you need to step back and get that right first before you start adding extra complexity to the mix by miking every last nut and bolt on the kit.
G.
Last edited:
R
Robert D
New member
Glen - this is a bit disappointing, and a bit insulting. I have given several "positive explanations" of why I, and the great majority of academia, call this phase reversal or phase inversion. It has been called this for decades, way before the first PN junction was created. Let's take this out of the hypothetical and deal with reality. I don't know where you came up with the notion that phase inversion can only occur around some arbitrary centerline, a different one for each component of a complex waveform. Please give your source for this? No matter the waveform frequency, complexity, shape, or offset, when it goes through a phase inversion, be it by passing through an NPN transistor in common collector mode, or through an inverting op amp, or by way of a transformer with a center tapped secondary, the resultant waveform is the antiphase counterpart of the input waveform. Each and every component of its waveform is 180 deg out of phase with the original. DC does not enter into this fact, nor does asymmetry change this fact. The two signals have a phase relationship that can only be described as 180 degrees out of phase.
If you describe how a push-pull amplifier works, part of that description is that the two sides are fed signals of opposite phase from the same source.
If you talk about an inverting Op Amp circuit, the output is described as being 180 deg out of phase with the input. Same with the Common collector transistor amp.
We could talk about a circuit called a paraphase amplifier.....Jesus Glen, I could stack textbooks to the ceiling, each one describing the simple concept of phase inversion, and citing it in a number of circuit descriptions!
I thought we had arrived at the point in this discussion where we were examining the finer points of its application in audio, not denying its existence and accusing me of "circular arguments", "emperor’s clothes", and other such dribble. The world is flat.....again.
And what is wrong with 9 page threads? That, to me, is what a discussion forum is all about, not the same questions over and over again that we all know the answer to. If the debate tires you, then you know what to do. But the scientist in you is betrayed by that kind of remark.
If you describe how a push-pull amplifier works, part of that description is that the two sides are fed signals of opposite phase from the same source.
If you talk about an inverting Op Amp circuit, the output is described as being 180 deg out of phase with the input. Same with the Common collector transistor amp.
We could talk about a circuit called a paraphase amplifier.....Jesus Glen, I could stack textbooks to the ceiling, each one describing the simple concept of phase inversion, and citing it in a number of circuit descriptions!
I thought we had arrived at the point in this discussion where we were examining the finer points of its application in audio, not denying its existence and accusing me of "circular arguments", "emperor’s clothes", and other such dribble. The world is flat.....again.
And what is wrong with 9 page threads? That, to me, is what a discussion forum is all about, not the same questions over and over again that we all know the answer to. If the debate tires you, then you know what to do. But the scientist in you is betrayed by that kind of remark.
noisewreck
New member
OK, so SSG had asked me to create a file that would have some asymmetry, while the RMS of the file was not "symmetrical", and some other criteria that I don't remember very well (it's 2:30am right now and I woke up at 6:30am the morning before)... I think it will satisfy.
The idea is that the waveform is skewed here-and-there, and the envelope has bumps and bruises.
As expected, Cubase's phase reverse creates and exact mirror of the original. When both versions are played together, the result is complete silence. Mathematically, it is as if the waveform has been multiplied by -1, which is what I suspected it would do all along.
So, if I read what Robert D says, phase reverse and polarity inversion are the same thing?
Dunno, and at this point don't care. I am gonna keep calling it polarity inversion, just because in my head, that's what it is.
Just chill now.
So, download the 24bit/48kHz file from YouSendIt.
Here's what it looks like in Cubase, with Phase Reversed version below:
The idea is that the waveform is skewed here-and-there, and the envelope has bumps and bruises.
As expected, Cubase's phase reverse creates and exact mirror of the original. When both versions are played together, the result is complete silence. Mathematically, it is as if the waveform has been multiplied by -1, which is what I suspected it would do all along.
So, if I read what Robert D says, phase reverse and polarity inversion are the same thing?
Dunno, and at this point don't care. I am gonna keep calling it polarity inversion, just because in my head, that's what it is.
Just chill now.
So, download the 24bit/48kHz file from YouSendIt.
Here's what it looks like in Cubase, with Phase Reversed version below:
Attachments
M
MCI2424
New member
You should download all the latest drivers for your video, soundcard, and motherboard. The popping and clicking are usually products of system interrupts. Video cards especially as they do thousands of re-draws every minute. Also, set your audio buffers to the acceptable size in your recording software and make sure you are using the proper windows audio driver.
M
MCI2424
New member
Here is a fact. The "Phase Reverse Switch" changes the polarity of the + and - signal conductors.
What it does to the sound is mathmatically provable by someone (probably a student at MIT).
I have no idea if you really can reverse the polarity of sound or not. But, sound is pressure through the air and time is the only real factor that can change sound (other than amplitude and frequency) so, it probably stands to reason that the reversal of electrical polarity causes a reversal of sound phase by 180 degrees. There are phase alignment devices that change the phase of the electrical signal anywhere between 0 and 180 degrees.
And, I don't think you can change the "phase" of a single mic anyway since "phase" is relational to something else, like another mic etc. But, in any case, I do agree with you that this subject has been beaten to death. I guess if you push the button and it sounds better, you should leave the button pushed?
SouthSIDE Glen
independentrecording.net
Wow, sometimes I just hate such prehistoric methods of communication as Internet forums; the level of misunderstanding can be so high. Rob, you took me in an entirely different way than I intended or could even imagine. Probably my fault for not embedding enough preemptive phrase couching in my post. I'm sorry of you felt offended by what I said, that's not how I meant it at all.
I have been continuing to think about this issue, and I think I can finally see how I have this wrong, but I'm not sure:
Let me ask you a question Rob: With simple phase inversion, how is the centerline/mirror defined?
If it is defined as the rest voltage - which it probably should be now that I think of it - then I take back everything I said and admit I was wrong to make the distinction between phase inversion and polarity inversion.
If, however, it is defined as the center of the voltage envelope of the wave, as I have been defining it all along - probably wrongly, now that I think about that - then everything I have said would be 100% true.
This is what sucks about the Internet. After pages and pages, and weeks of back and forth, nobody was able to catch that flaw in my definition.
That's all I was asking for: "Where was I wrong?" was my question. Am I really THAT bad at explaining myself?
Well, I'm not sio bad where I can't admit a mistake once I do see where my thinking was wrong, and where I can't apologize for dragging so many people through so long of a process just so I can get it right in my head, and where I can't thank you all for helping in your well-intentioned ways.
I made a basic mistake in my understanding. I was wrong. I apologize for taking so long to realize how I was wrong and correct myself. And I thank everybody involved to helping out and putting up with my slowness on the uptake.
Now, if you'll excuse me, I'll go get some milk to wash down this humble pie
.G.
R
Robert D
New member
Oh man......now I feel bad. 
My pen was a bit sharp yesterday, and yes, this is an often misenterpreted means of communication. No humble pie is necessary Glen, this is the way science is properly advanced. If no one rigorously attacks and attempts to disprove each and every hypothesis, we end up with bad science. No one was dragged through anything, unless they have a compulsive disorder that causes them to click on and read through threads against their will.
Regarding your question; yes, I think you phrased that perfectly.....the centerline of the mirror would be the rest voltage, be that 0vdc in a balanced circuit, or the middle of the linear portion of the transfer curve in a single ended circuit.
In the end, it really doesn't matter what people call it, polarity inversion or phase inversion, as long as they understand that the resulting phase relationship is truly 180 deg out of phase with no shift along the x axis. Without this truth, many of the tools we have in our audio toolbox disappear.
My pen was a bit sharp yesterday, and yes, this is an often misenterpreted means of communication. No humble pie is necessary Glen, this is the way science is properly advanced. If no one rigorously attacks and attempts to disprove each and every hypothesis, we end up with bad science. No one was dragged through anything, unless they have a compulsive disorder that causes them to click on and read through threads against their will. Regarding your question; yes, I think you phrased that perfectly.....the centerline of the mirror would be the rest voltage, be that 0vdc in a balanced circuit, or the middle of the linear portion of the transfer curve in a single ended circuit.
In the end, it really doesn't matter what people call it, polarity inversion or phase inversion, as long as they understand that the resulting phase relationship is truly 180 deg out of phase with no shift along the x axis. Without this truth, many of the tools we have in our audio toolbox disappear.
SouthSIDE Glen
independentrecording.net
Yeah, my fundamental mistake was defining the centerline of an asymmetrical wave incorrectly. That incorrect definition is what was producing the offset that I was seeing.
Never mind!
G.
kubeek
New member
Yeah, my fundamental mistake was defining the centerline of an asymmetrical wave incorrectly. That incorrect definition is what was producing the offset that I was seeing.
Never mind!
G.
Well I think it was me who took you in that direction.. I was saying that the center line is in the middle of peaks, but sadly that is only true with sine signals that don´t change in time.
But I´m glad that we have this battle of phase vs polarity inversion sorted
SouthSIDE Glen
independentrecording.net
Not your fault, kubeek. This debate has been ongoing since before you joined this board (lucky for you, unlucky for meWell I think it was me who took you in that direction.. I was saying that the center line is in the middle of peaks, but sadly that is only true with sine signals that don´t change in time.
But I´m glad that we have this battle of phase vs polarity inversion sorted
).Actually I am going to back up just a little bit upon even further thought (oh, boy, here we go again...
)My original assertion that all it took was asymmetry to differentiate between phase and polarity was wrong, for the reasons just explained. I reamain wrong and corrected on that point.
However, under the special condition where there is offset going into the inversion, there is still a difference between the two processes.
Yes, they both represent a mirroring around the rest voltage. But if there is DC offset going in, that indicates that the rest voltage is no longer 0DC, therefore there is a Y axis shift in the placement of the mirror for phase inversion, while the mirror for polarity inversion remains the same.
Phase inversion and polarity inversion, in and of themselves still remain different animals, even though that difference only reveals itself in special cases.
Rob was right earlier, that anybody with a head on their shoulders would perform a DC offset rectifcatoin before flipping, since there is rare if ever a desire to keep that offset anyway, in which case then the end result would remain the same. However they definitions remain different.
To correct what I said earlier:
1. Polarity inversion is a mirroring of the waveform around 0DC.
2. Phase inversion is a mirrorig of the waveform around the waveform's rest voltage.
3. When the waveform's rest voltage is 0DC, phase inversion and polarity inversion are the same thing.
Does that sound about right?
G.
R
Robert D
New member
In all cases, the phase relationship between all AC components of the waveform from input to output is 180 degrees out of phase. So, all of these cases result in phase inversion. That fact is independant of the DC component of the signal. But the DC component needs to be resolved before anything can be done with the antiphase signals in terms of functions like common mode rejection. Even if just the antiphase output is being used, the next stage won't be happy if the signal has a large DC component. That's where de-coupling comes in, blocking the DC component and passing on only the AC component.
In practical terms, the only polarity reversal that happens is in balanced signals, where hot and cold are swapped. This is fine because they share the same reference to gnd, and everything stays happy. You can't just polarity reverse a single ended signal, you have to either use an active device or a transformer to affect the inversion. The transformer blocks the DC component, and the active design is usually decoupled and biased properly so that this is not an issue.
If anyone has schematics to post of equipment with phase invert function, please post it so that we can look at some real world cases.
noisewreck
New member
We do? So, then what's the difference between polarity inversion and phase inversion when they're essentially one and the same. And if they are the same, why have two different terms for the same thing? ARGH!But I´m glad that we have this battle of phase vs polarity inversion sorted
No wonder they don't teach evolution in Kentucky anymore. Sometimes it's easier understanding a petty and angry god than all this scientific mumbo-jumbo.
SouthSIDE Glen
independentrecording.net
A big part of our disagreement, Rob, (as I see it) is that I'm talking theory and definition, and you're talking design and circuitry. We are both right in our own domains, I think.
The fact that any DC issues are filtered out of the equation by the circuit design, and therefore polarity switching and phase switching always have the same result, which is - correct me if I'm wrong - the core of your case does not - at the core of my case - change the fact that the description, definition and mechanism of phase inversion and polarity inversion are indeed different.
In fact, I'd say that your description actually proves my point. What you're saying is that the DC issues "need to be resolved" before the polarity can be ensured to do as desired without potential issues downstream. This goes back to what I was saying in the last thread. In that view, you're right, phase inversion indeed always end in the same result as polarity inversion because any chance for offset is removed first.
But what that really says is that DC recification plus phase inversion equals polarity inversion. It does not say that phase inversion equals polarity inversion. Like I said last time, just because (3x2)-1 = 3+2, and 1 is always subtracted from the first term in real life, that doesn't mean that 3x2 = 3+2.
Put the schematics down for a second Rob, and explain to me how in a situation of DC offset that phase inversion is identical to polarity inversion. Explain where that original diagram I had of the three sine waves is incorrect. Put another way, pick up the math and show me how the math describing both of the processes is results in the same solutions regardless of the input values. Because I just don't see how that is possible.
Polarity inversion is simply a change in sign. Phase inversion is not. Phase inversion is a change in relative value, measured in degrees. No sign. Tell, me, is 180° positive or negative? If the change is instantaneous and not a shift, then it's neither.
The fact that by man-made design the input voltage values are purposely constrained and insured to conform to favorable values that make polarity inversion look identical to phase inversion does not make them the same...not any more than automatically subtracting 1 from 6 to make 5 means that 6 is equal to 5.
G.