[Russell Dawkins]

For example if you pluck a string towards you you'll hear one polarity, and if you pluck it the other way you'll hear the opposite.

A strange choice of examples. If you are talking about, say, a guitar then the first impulse will be either a rarefaction or compression, depending on the listening position, as the major source of the sound heard is not the string but the upper surface of the sound box. This is regardless of which way the string is plucked since what is imparted to the instrument from the string is variations in tension caused by the vibration. The lateral motion of the string doesn't count for much.

Same thing if you strike a single membrane or gong.  Depending on where the mic/ear is, the leading edge of the signal can be either a rarefaction or a compression.
Secondly, when you mix channels from multiple mics you often have to delay or invert the polarity of some of them to avoid phase cancellations.  Actually mics routinely have a polarity inversion switch for that kind of thing.

Actually, I have never seen a mic with a polarity inversion switch. Mic preamps and channel strips on mixing boards usually have them.
 
So by the time the recording gets mixed down to two tracks, any notion of a correct polarity is gone.
 
Not true in my recordings, and I am not alone. I am aware of the polarity of the signal all the way through every single stage (and there are many) of the recording /mixing/mastering process. I would never buy or use a piece of equipment without knowing whether it inverted polarity. I have come across two cases where it was a real factor in my experience - both about 12 years ago. One involved a piece (spl Vitalizer) intended as a mastering tool. I began to notice that it always sounded better in some way when the box was out of the circuit versus "hard wire" bypass button engaged. Then I did a little careful comparison, putting the signal simultaneously through the box and through two other channels on the board. Bringing up the levels on both pairs to match revealed that "bypass" inverted polarity.

The other involved the Tascam DA30, a popular DAT recorder that small studios used to mix down to in the early days. It's analog balanced inputs were pin 3 hot, whan most of the industry was pin 2 hot. This meant that any CD made from DAT masters made on this machine were absolute polarity-inverted, even if care was taken everywhere else in the recording chain to preserve correct polarity. In subsequent models Tascam, who was a pretty big player in the studio world, quietly reversed the polarity of their ins and out to pin 2 hot.

[opaqueice]

A strange choice of examples. If you are talking about, say, a guitar then the first impulse will be either a rarefaction or compression, depending on the listening position, as the major source of the sound heard is not the string but the upper surface of the sound box. This is regardless of which way the string is plucked since what is imparted to the instrument from the string is variations in tension caused by the vibration. The lateral motion of the string doesn't count for much.

But I didn't say a guitar, I said a string.  A guitar is more complicated.  In any case it sounds like we agree that the polarity depends on position, which was what I wanted to illustrate.

Actually, I have never seen a mic with a polarity inversion switch. Mic preamps and channel strips on mixing boards usually have them.

Picky picky... maybe I'm misremembering, but I'm pretty sure I have.  Anyway it's not very relevant where the switch is, just that it exists.

Not true in my recordings, and I am not alone. I am aware of the polarity of the signal all the way through every single stage (and there are many) of the recording /mixing/mastering process. I would never buy or use a piece of equipment without knowing whether it inverted polarity. I have come across two cases where it was a real factor in my experience - both about 12 years ago. One involved a piece (spl Vitalizer) intended as a mastering tool. I began to notice that it always sounded better in some way when the box was out of the circuit versus "hard wire" bypass button engaged. Then I did a little careful comparison, putting the signal simultaneously through the box and through two other channels on the board. Bringing up the levels on both pairs to match revealed that "bypass" inverted polarity.

Being aware of the polarity doesn't suffice.  How do you deal with mixing channels from mics at different locations?  Do you make preserving polarity a goal?

For example, suppose you want to record a single acoustic guitar.  I imagine you want more than one mic so that you can mix and adjust the tonal balance.  But then to avoid phase cancellations, depending on the mic positions, you may want to invert polarity, no?

I'm curious - what's your opinion on the audibility of polarity reversal?

[Russell Dawkins]

Picky picky... maybe I'm misremembering, but I'm pretty sure I have.  Anyway it's not very relevant where the switch is, just that it exists.

It is relevant to your presumption of expertise on the subject.

Not true in my recordings, and I am not alone. I am aware of the polarity of the signal all the way through every single stage (and there are many) of the recording /mixing/mastering process. I would never buy or use a piece of equipment without knowing whether it inverted polarity. I have come across two cases where it was a real factor in my experience - both about 12 years ago. One involved a piece (spl Vitalizer) intended as a mastering tool. I began to notice that it always sounded better in some way when the box was out of the circuit versus "hard wire" bypass button engaged. Then I did a little careful comparison, putting the signal simultaneously through the box and through two other channels on the board. Bringing up the levels on both pairs to match revealed that "bypass" inverted polarity.

Being aware of the polarity doesn't suffice.  I thought that all that was needed!! 
How do you deal with mixing channels from mics at different locations? Oh, I just throw them all in together.   
Do you make preserving polarity a goal? Nah - I can't be bothered.

For example, suppose you want to record a single acoustic guitar.  I imagine you want more than one mic so that you can mix and adjust the tonal balance. 
Well, you're wrong. If I use more than one mic, it's to capture the instrument in stereo, not so I can play with it later to conjure up a "tone".
But then to avoid phase cancellations, depending on the mic positions, you may want to invert polarity, no?
No, but depending on the situation, I will want to make sure it's in the same polarity as the room mic, then apply the appropriate delay so the arrival time in the mix is the same as the room mic. Keep in mind that if this is not done and the room mic is, say 12 feet from the musician and his spot mic, then around 12 msec of delay is required on the spot for there to be no smearing of the image. Also keep in mind that here we are talking about relative polarity, where the subject of the thread originally was absolute polarity.

I'm curious - what's your opinion on the audibility of polarity reversal?

Relative polarity is very obvious; audibility of absolute polarity is easier on simpler playback systems, like crossoverless wideband driver-based speakers, and with simpler (fewer mics) recordings. Even then it can be subtle, but perceptible, if not blatantly obvious as Clark Johnson suggests.

[opaqueice]

Whoa there...  I never claimed expertise as a recording engineer, so I'm not sure why you would presume I have any.  I have a decent knowledge of electronics and know more than a little physics, but that's as far as I go.

I'm a little confused by your response - if polarity is so subtle why not use several mics and then mix down?  Is there some other reason you prefer not to do that?

As for time delay, that makes sense - except that as we've agreed, if the mics are at different locations they may "hear" a different polarity anyway.  Furthermore if there's more than one source of sound I don't see how there can be any choice of time delay that preserves coherence for more than one of the sources.  Am I missing something there?

[Russell Dawkins]

Whoa there...  I never claimed expertise as a recording engineer, so I'm not sure why you would presume I have any.

No offense, but I didn't, actually.

I have a decent knowledge of electronics and know more than a little physics, but that's as far as I go.

I'm a little confused by your response - if polarity is so subtle why not use several mics and then mix down? 

I think I may be missing your point here. We are talking about absolute polarity, aren't we? I do use several mics when I have to, but for me absolute polarity is not part of the equation. Proper relative polarity and time alignment is. All of my mics are pin 2 hot except for some Russian ones. Whenever I use those, I flip phase on the preamp. On rare occasions I will flip phase on drum overheads depending on their location in relation to the kick drum skin.

Is there some other reason you prefer not to do that?

I don't not do that. I use several mics whenever there is an advantage.

As for time delay, that makes sense - except that as we've agreed, if the mics are at different locations they may "hear" a different polarity anyway.
I don't see it quite like this, but I'm sorry, I don't have the inclination to thrash through this, so can we let this point slide?

Furthermore if there's more than one source of sound I don't see how there can be any choice of time delay that preserves coherence for more than one of the sources.  Am I missing something there?

No, you're not but usually the need is only for an approximation. Anything within ± 5 msec is fine when the spots are only lightly used and the group being recorded is physically large, like an orchestra, large choir, big band and such. Beyond 10 msec, with situations where the spot is representing a particular instrument in thee mix at levels similar to those for the same instrument in the main mic (which I have for convenience been calling a room mic), the dual attack nature of the sound starts to become apparent. At 15 msec it is obvious, if there is any percussive aspect to the sound - say vibes, piano, xylophone, for example.

Strictly speaking the only way to have phase coherence for all instruments is to use a coincident stereo pair to record them even if the ensemble is huge, which is why I did just that, against all odds, in 1994 when I recorded the Ukrainian Radio and Television Orchestra in Kiev. Here was a 95 piece orchestra being recorded with a single pair of ribbons, with the conductor's enthusiastic support (after many lengthy discussions of just the type we are having here). A couple of equipment and music reviewers I know are using this recording as a reference for recorded orchestral sound so theory seems to fit reality here.

[Housteau]

It has been a while since I had started a thread which had evolved into such a lively subject.  I have learned a lot by following it.  It also reminds me a bit of what I do in my free time .  Just check my gallery photos .

[Double Ugly]

A couple of equipment and music reviewers I know are using this recording as a reference for recorded orchestral sound...

And at least a few of us non-reviewers, too. 

[opaqueice]

It is relevant to your presumption of expertise on the subject.

Whoa there...  I never claimed expertise as a recording engineer, so I'm not sure why you would presume I have any.

No offense, but I didn't, actually.

None taken, but I still don't get it....  anyway, let it drop.

As for time delay, that makes sense - except that as we've agreed, if the mics are at different locations they may "hear" a different polarity anyway.
I don't see it quite like this, but I'm sorry, I don't have the inclination to thrash through this, so can we let this point slide?

As you want.

Furthermore if there's more than one source of sound I don't see how there can be any choice of time delay that preserves coherence for more than one of the sources.  Am I missing something there?

No, you're not but usually the need is only for an approximation. Anything within ± 5 msec is fine when the spots are only lightly used and the group being recorded is physically large, like an orchestra, large choir, big band and such. Beyond 10 msec, with situations where the spot is representing a particular instrument in thee mix at levels similar to those for the same instrument in the main mic (which I have for convenience been calling a room mic), the dual attack nature of the sound starts to become apparent. At 15 msec it is obvious, if there is any percussive aspect to the sound - say vibes, piano, xylophone, for example.

Strictly speaking the only way to have phase coherence for all instruments is to use a coincident stereo pair to record them even if the ensemble is huge, which is why I did just that, against all odds, in 1994 when I recorded the Ukrainian Radio and Television Orchestra in Kiev. Here was a 95 piece orchestra being recorded with a single pair of ribbons, with the conductor's enthusiastic support (after many lengthy discussions of just the type we are having here). A couple of equipment and music reviewers I know are using this recording as a reference for recorded orchestral sound so theory seems to fit reality here.

Interesting, and very informative - thanks.  Occasionally I've listened to binaural recordings, and through a laptop soundcard and mid-fi phones they sound so realistic it's literally scary - more so than all but the most carefully set up ultra hi-end stereo systems.  I guess there must be a reason for that.  I never could believe it was just phase and polarity, but proper time alignment of sudden sharp sounds makes a lot of sense.

I'll look for your recording - I'd love to hear it.

[Daygloworange]

I'll look for your recording - I'd love to hear it.

 Russell's stuff is good. 

 Check it out for sure. It's worth it.

Cheers

[Russell Dawkins]

Interesting, and very informative - thanks.  Occasionally I've listened to binaural recordings, and through a laptop soundcard and mid-fi phones they sound so realistic it's literally scary - more so than all but the most carefully set up ultra hi-end stereo systems.  I guess there must be a reason for that.  I never could believe it was just phase and polarity, but proper time alignment of sudden sharp sounds makes a lot of sense.

I'll look for your recording - I'd love to hear it.

It may sound trite and simplistic, but I think the fact we do have only two ears is a clue.

Can you imagine trying to make sense out of the input from two big ears and 22 little ones scattered throughout the orchestra, even if you had the organic equivalent of a mixing board in your cranium?

I believe simpler tends to sound better from one end of the process to the other - microphone to speaker.
Fewer mics, fewer gain stages, conductors, contacts, plugs, ICs, capacitors, transformers, A/D, D/A converters.
I have heard that purely mechanical recordings can sound shockingly good - ones where the signal has never been through any electronics.

I think that in the world of two channels, the standout systems as far as spatial realism are concerned are, for headphones (and particularly in-the-ear types),  binaural and for speaker reproduction, Blumlein. Blumlein, to be properly played though, must be played through speakers with an included angle of 90º, not 60º which is the stereo standard.

I enjoy collecting signatures from forum posters and I saw one just now on another forum which relates to all this, I think:

The difference between theory and practice is that in theory there is no difference between theory and practice, but in practice there is.

[Russell Dawkins]

another one I like is:

"If you notice the sound, it's wrong"

I actually use this when mixing my stuff.

[Ethan Winer]

Do you agree Steve/Ethan?

1.   Absolute polarity can be audible under very specialized test situations.    Audible says nothing about better/worse, it just means its detectable.   

2.  Under the vast majority of situations its not audible.

Yes, I agree with that. It's not a big deal. Especially when compared to stuff that really matters a lot like room acoustics that almost everyone ignores completely.

--Ethan

[Ethan Winer]

I think you're forgetting something - ears are connected to brains, and brains do a lot of processing.

I don't know what that means in practice. What really matters is if the sound is changed enough to affect your enjoyment. To me it's a non-issue, even for sounds with sharp leading edge transients like a kick drum. It seems like there'd be a big difference in the sound if the drum head (or loudspeaker) goes in rather than out. But in practice it makes little if any difference.

Using a PC ABX program, I listened through Grado headphones to the two tones. The effect wasn't obvious, but listening carefully you could hear a slight difference in pitch. Let's say A sounded higher than B, and I could pretty easily score perfectly on the ABX by listening for it. So the polarity reversal was audible, but was it my ears or was it the headphones or something else in the playback chain?

Your observation "wasn't obvious" is the key.

I don't know if it's your ears or a nonlinearity in the headphones. It may depend on the frequency components present, as I explained earlier about the sawtooth wave. In that case the reversal was audible at 20 Hz but not at 50 Hz. BTW it was the same with headphones or with speakers with the sawtooth wave.

Again, I won't argue against the importance of absolute polarity too strongly because it apparently is discernible in some cases. But I have never seen a situation where it makes a meaningful difference in sound quality. It seems more a curiosity than anything else.

--Ethan

[opaqueice]

I don't know what that means in practice. What really matters is if the sound is changed enough to affect your enjoyment. To me it's a non-issue, even for sounds with sharp leading edge transients like a kick drum. It seems like there'd be a big difference in the sound if the drum head (or loudspeaker) goes in rather than out. But in practice it makes little if any difference.

Well one thing's for sure - if we were (consciously) aware of phase information we'd go crazy from information overload, since every time we moved our ears at all it would change drastically.

I don't know if it's your ears or a nonlinearity in the headphones. It may depend on the frequency components present, as I explained earlier about the sawtooth wave. In that case the reversal was audible at 20 Hz but not at 50 Hz. BTW it was the same with headphones or with speakers with the sawtooth wave.

Did you read the rest of my post?  If it was nonlinearity in the phones it wouldn't have reversed when I turned them around.  Therefore it was NOT in the phones.

Again, I won't argue against the importance of absolute polarity too strongly because it apparently is discernible in some cases. But I have never seen a situation where it makes a meaningful difference in sound quality. It seems more a curiosity than anything else.

Agreed.

[Ethan Winer]

If it was nonlinearity in the phones it wouldn't have reversed when I turned them around. Therefore it was NOT in the phones.

That's too complicated for me to ponder.

Seriously, why would it NOT reverse? Or do I not understand what you mean by reverse?

[opaqueice]

That's too complicated for me to ponder.

Seriously, why would it NOT reverse? Or do I not understand what you mean by reverse?

It is a little mind-bending sometimes  .  But I think we can sort it out.

Suppose the effect were in the phones, meaning that they respond slightly differently to positive voltages than to negative.  There are a lot of reasons why that might happen, but let's not worry about the "why" for the moment - just suppose it's the case.  Then when you play a signal with more power in the positive half it will sound different than one with more power in the negative.  Turning the phones around doesn't change the voltage and therefore doesn't affect this.*

Now suppose instead the effect is in your ears, meaning they respond differently to compressions than the rarefactions.  If so, audio signals with more power in positive voltage will sound different than the opposite, because they will cause the speakers to generate sound waves with more power in compressions than rarefactions (or the other way around, depending).  But now the key point is that if you have an open-baffle speaker (or a single open-backed headphone driver) playing such a sound, the polarity will be reversed if you rotate it 180 degrees.  In other words when the cone is moving towards your ear it makes a compression, but when it moves away it makes a rarefaction.  So rotating the headphone speaker turns rarefactions into compressions. 

When I did this experiment, rotating the speaker interchanged the pitch - if A > B before, after rotating B > A.  That proves the effect has nothing to do with the electrical part of the system, so it's got to be in my ears.

Make sense?


*Caveat - one possible source of asymmetry in sealed-box speaker drivers is the cushion of air behind them and in front reacting differently to compressions versus rarefactions.  However an open-backed headphone doesn't have such a cushion.

[Ethan Winer]

Make sense?

Maybe it would if I could see the phones and how they implement the reversing. But just being open back is no assurance there's no back pressure unless the back is totally open with no support vanes etc.

Can you post a link to a photo of the phones or a product description?

--Ethan

[Russell Dawkins]

*Caveat - one possible source of asymmetry in sealed-box speaker drivers is the cushion of air behind them and in front reacting differently to compressions versus rarefactions.  However an open-backed headphone doesn't have such a cushion.

Actually it does, except that it is between the diaphragm and your ears! (unless it's an AKG K-1000 or similar)

I don't get your drift at all regarding the rest of the your last post.

When you say "turning the phones around" do you mean reversing them on your head - left for right?

[opaqueice]

Maybe it would if I could see the phones and how they implement the reversing. But just being open back is no assurance there's no back pressure unless the back is totally open with no support vanes etc.

That's true, but the only relevant thing here is whether reversing them could _change_ the back pressure, and it can't.

See below for a link.

Actually it does, except that it is between the diaphragm and your ears! (unless it's an AKG K-1000 or similar)

Yes, I worried about that - but even with the phones held away from my ears the effect remained, but with them turned around it switched.

When you say "turning the phones around" do you mean reversing them on your head - left for right?

No.  Left and right are irrelevant here - you could do this test with just one channel active using one ear.  I mean you rotate the cups at the ends of the head band around the axis defined by the band, so that (if you rotate by 180 degrees) the speakers end up pointing out away from your head.   

Here's a pic of some similar ones - you can rotate that silver rod freely, so if you rotate it by 180 the speakers point out.

http://arstechnica.com/images/headphones/sr80-full.jpg