Moulton Laboratories
the art and science of sound
About Comb Filtering, Phase Shift and Polarity Reversal
By Dave Moulton
August 1993
2. "Polarity" and "Phase"

The phenomena of comb filtering, phase shift and polarity reversal are surrounded in myth and are a bedrock topic for audio engineers.

< 1 2

"Polarity" and "Phase"

All this, so far, is straight-ahead no-brainer stuff, but it now begins to get a little weird. Two phenomena that we have to ponder out on the edge of the audio twilight zone are "the quality of out-of-phase-ness" and "how come we don't always hear out-of-phase-ness."

If you think about it, it's obvious that there is no such thing as "out-of-phase" for a sound that has more than one frequency. As the figures above illustrate, any given delay time will yield an infinite number of phase shifts at an equally infinite number of different frequencies. So how can a sound be 180° out of phase?

Answer: it can't, unless it's a sine wave.

But, Dave, there's an out-of-phase button on my console that can be used for some killer effects, and we're not talking sine waves.

Answer: what you are referring to isn't "out-of-phase," it is "reversal of polarity."

Yeah, so how come it is called "out-of-phase?"

Answer: because when you reverse the polarity of a sine wave, the sine wave appears to have shifted in time by 180° (even though it hasn’t), and electronics engineers, like everyone else, are lazy and sloppy with their language. And, like, they know what they mean, you know?

So Dave, what do you mean by "reversing the polarity?"

Answer: reversing the polarity means reversing the positive and negative portions of the wave (in electricity it means reversing the direction of current flow, in air it means changing compression to rarefaction and vice versa, and in math it means changing the signs from positive to negative and vice versa).
  
A signal and its inversion. The inversion is offset above the original for clarity of illustration. In reality, they both swing back and forth across the same baseline or zero-point.

When you reverse the polarity of a signal, there is no time change, hence there is no phase shift. Also, polarity reversal has some special behavioral characteristics of its own.

This leads to the second big issue: how come we don't always hear “out of polarity?”

C’mon, admit it. You don’t.

Think about it. Phase shift and comb filtering are everywhere. In a typical room you are usually listening to several hundred delayed versions of original signals all the time, and every time you shift your head by a couple of inches, you are into big-time acoustical phase shifting and comb filtering. But you don’t really hear them, even though they are predominant in the environment. If you don’t believe me, run down to your local Radio Shack, pick up their $35 Sound Level Meter (one of the great audio bargains of our century and a necessity for every truly cool recording engineer), run back to your studio, put up a 2 KHz. sine wave in the monitors, stand back in the room and start observing the levels on the Sound Level Meter. As you slowly move the meter around, the levels will swing wildly across the range of the meter. These varying levels are due to the constructive and destructive interference patterns caused by changing phase shift between the direct signal and all the reflected paths as the meter’s location changes. But when you move your head around, you may hear a little change in level, but hardly any sense of phase shift at all -- certainly not like negative killer flange or other bozo effects so dear to us all.

The truth is that our magnificent hearing system doesn’t present the state of phase of sound waves to our conscious mind. It detects phase all right, but uses it to help with localization instead of timbre. We use all that phase shift data to localize sounds in space, but we don’t consciously detect it as a sound quality.

Except when it all comes from the same point in space.

I’ve talked about this in other articles, but it is worth going over again. When you sum two signals electrically, they become one signal, acoustically speaking. When we radiate that combined signal into the room, our ears detect it as the source and detect all the room reflections (which come from other directions) as delayed versions. But if that complex mixed signal has phase shifting or comb filtering going on, it affects the frequency response (and timbre) of the original source and we hear it, big time.

So what does it all mean? How can you use this all for fun and profit? A couple of things:
  • The comb filtering that occurs in rooms as a result of reflected sounds in the room is pretty benign stuff. It helps you localize sound sources and figure out the room. It usually doesn’t affect the perceived timbre of the sound coming from the loudspeaker very much. Only after you’ve got most of the other acoustic issues under control will you want to worry about the effect of comb filtering between loudspeakers and reflections in your mixing room.
  • Phase shift is one of the three fundamental attributes of sound waves (the others are frequency and amplitude). It is a time-based phenomenon that is at the root of comb-filtering and our ability to localize sound sources in rooms.
  • The term “180° out-of-phase” is a misnomer for all sounds except sine waves. Mostly we mean “out-of-polarity” instead. Polarity reversal is unrelated to time. Interesting, the high-tech item here is polarity reversal (“PolRev,” for short). Phase shift occurs all the time in non-technological nature, but the manipulation of polarity of a sound (in either the acoustical or electronic realms) exists only in high-tech items such as loudspeakers (acoustically) and audio circuits (electronically). The audio problems pertaining to comb-filtering and polarity reversal are due to these technological developments and the fact that our hearing mechanism didn’t evolve in preparation for such high-tech oddities as stereophonic sound.
  • Getting your speakers in polarity is essential. This includes getting each driver in a speaker system in polarity with its equivalent driver in the other system in the pair.
  • Keeping track of polarity in your microphones, mic cables, console(s) and other audio gear is essential. This isn’t too big a problem with unbalanced semi-pro stuff, but balanced lines (which normally use XLR connectors) can easily and inadvertently be wired out of polarity. Always make sure that pin two is wired to pin two, etc.
  • When checking the monaural quality of a stereo mix, comb filtering that wasn’t at all apparent in stereo may show up big time. Have an ear out for it, and if you hear it, you may have serious phase shift and/or polarity problems to fix, as well as possibly some polarity problems. How you actually fix them may be as easy as flipping the polarity button on a module on your console, or it may involve microphone or other adjustments that are well beyond the scope of this article.
In summary, this is one of those bedrock topics that isn’t particularly obvious when you start trying to record your music. What makes it particularly tricky is that it’s only audible some of the time, so you have to listen sharp. Even when you can’t hear it, it can still cause problems in your recording. On the plus side, comb filtering, phase shift and polarity reversal are all phenomena that can really help you easily and professionally solve a variety of audio problems, particularly stereo audio problems, once you get a handle on how to work with them. It’s worth spending the time thinking about, experimenting with and listening to them. Then you too can use them for fun and profit!

Dave Moulton is Chairman of Music Production and Engineering at Berklee College of Music in Boston. Alex Case is a student there. Peter Alhadeff teaches math there.
< 1 2

COMMENTS

Kansas City, MO     Aug 25, 2006 12:17 AM
So why is it that my turbosound TMS-3 rig calls to invert the polarity of the mid and high drivers? I thought it was because of the phase problems of analog crossovers but when i spoke with them they said that I could achive the same results by wiring my system in phase at all points and placing a 1ms delay on the mids and highs. So now that we all have digital crossovers should everything be wired in phase and not delayed to preserve the phase coherent design of the box or should we use the delays availible in modern crossovers to more acurately time-allign the box OR should it be left alone and wired out of phase and not delayed? I suppose my question is why were they designed to run mid/high out of phase and would that change when switching to a 4 way system.
Ryan Chamberlain 
     Aug 25, 2006 11:13 AM
Your initial assumption is, so far as I know, correct.

The inverting of polarity is intended to reduce destructive interference at the crossover frequencies, while ignoring the compartively minor issues of polarity in the driver's passband.

I wouldn't be so casual about simply applying a 1 ms. delay. However, if you are triamping and have DSP/xovers prior to the amps you can certainly do a nicer job of cleaning up the driver interactions than can be done with passive crossovers. To get it relly good, you are going to need some pretty good measurement instrumentation.

As for the last part of your question, I don't know why they were designed as they were, or what assumptions they are making. Unfortunately, it sounds like they aren't being entirely clear with you.

I hope this helps.

Dave
DaveMoulton 
Budapest, Hungary     Jan 15, 2007 09:13 AM
Hello! I've found in some articles that phase shifting in EQ'ing can really affect the attack of the sound... Having known this, I've got to ask whether it could be a problem when you are setting up your amplifiers at a concert hall or a rehearsal room. I'm a guitarist and I started to wonder the thing I've mentioned before.
There are tips and tricks to avoid that phase shifting effect (using broad Q's for boosting, using only a small amount, and cutting with narrow Q's) in mixing, that I know, but is this useful for a rock band, setting up their amplifiers? (2 guitars, 1 bass)
Andras Toth 
     Feb 08, 2007 03:39 PM
For anyone's interest, a similar article on this topic can be seen over at Whalco Development:

http://www.whalco.ca
Webmaster 
     Jul 13, 2007 04:33 AM
In your tutorial explained polarity change of one of srereo signal and given that to chnge the pola rity if one channel for the solution for this If same frequency components in both channels this is ok But if same frequency component not totally in both channel this solution some time give wrong solutions. But phase correlation meter indicate phase out please give explaination about this also from your tutorial
Asoka 
Groton, MA     Jul 17, 2007 01:33 PM
I'm not sure I completely understand your comment/question. Any "difference" components in a stereo signal will NOT be cancelled when polarity is inverted for one channel and the signals are summed to mono, while all the identical components will be cancelled. See my pieces on A+B/A-B listening and analysis techniques.

The phase correlation meter simply indicates the relative amplitude of the "sum" and "difference" components (the in-polarity vs. out-of-polarity parts) of a stereo signal. It's a simplified view of what is going on in a stereo signal that has some limited usefulness.

I hope this helps.

Best,

Dave
Dave Moulton 
     Sep 22, 2010 03:18 PM
Dear Dave,

FYI at the following link:  http://www.ultrabitplatinum.com/?page_id=725   .  If you find that interesting, I can send you a text only version sans commercial references.  You may also find my think piece about polarity at http://www.ultrabitplatinum.com/?page_id=88 interesting as well, or not. 

If you'd help me sort out the mistakes, when and if they're corrected, you just might become an instant overnight music-lovers' folk hero. Or on the other hand be tarred and feathered and have your ears cut off and served up to you on a silver platter for your last meal as an audiophile guru.  I invite you to visit so that I can take you out for a meal on me and then we can share some of your music and mine on my custom built (by me) audio system on a CD player with a digital domain remote controlled polarity switch.  My system is the best I've heard so far for discerning polarity which still must be done by ear. Those who hear CDs played over my system don't usually have much trouble hearing polarity.  But whatever happens it will be just between the two of us, if that's what you prefer. It'll be attorney client privilege, or Perfect Polarity Pundit privilege, you decide.

Besides email may also call me 7 days a week at 619-401-9876 or toll free at 888-588-9542 between 9AM and 11:45PM Pacific Time.

Best regards.

George S. Louis, Esq., CEO
Digital Systems & Solutions
Email: gslouis@gslouis.com
Website: www.UltraBitPlatinum.com
Phone: 619-401-9876
1573 Kimberly Woods Drive
El Cajon, CA 92020-7261

P.S. How difficult would it be for companies' voice menus instead of saying "Please listen to the entire announcement because our menu has changed recently before making your selection" to "Our announcement was last changed on such and such date?
georgelouis 
France     May 30, 2013 03:38 PM
This is great ! I've been looking for this for ages.
Thx again for the explanation.
capillotraktor 

Post a Comment



rss2

rss atom