|Max dB with Doug Blackburn
Back Issue Article
Audio Signal Polarity A Look Behind the Curtain
There has been a lot of discussion over the years about the audibility of audio signal polarity reversal. Some claim it is obvious, and others claim it is inaudible. How can this be? Either something is there or it is not, unless the people who believe they can hear polarity reversal were not conducting their tests correctly. Whenever you select a test or evaluation method, there is always a possibility that you will make a choice in setting up the test that will contaminate the results due to unexpected or unforeseen variables being introduced. These variables can actually have more influence on the outcome than the one parameter that was supposedly being evaluated. This is a major pitfall in scientific research. The scientist selects a test with hidden variables and the test ends up being invalidated later when it is realized that the test or experiment introduced variables that were not anticipated.
Lets define audio signal polarity first. Music, when it exists as an electrical signal, is "alternating" in nature. In most audio circuits, the signal is positive some of the time (50% on average) and negative some of the time. When sound exists in the air, there is no positive and negative -- there are sound waves which travel as compressions (higher pressure) and rarefactions (lower pressure). Your eardrum moves inward on compressions and outward on rarefactions. It would stand to reason then that if you reverse the polarity of the audio signal in the electrical domain, you would reverse the order in which your eardrum would receive compressions and rarefactions and that this would then be clearly audible. Reason makes this connection seem inevitable and irrefutable. To think otherwise would seem impossible and downright dumb. Well, if you like the underdog, youll love this subject. In spite of the apparent logic of the situation, there are excellent reasons for all that youve read or "know" about the reversal of audio signal polarity being audible to be incorrect. The experiments that prove it is audible have been incorrectly performed for decades and the human hearing mechanism apparently does not operate as everyone assumes.
Good intentions, bad experiments
The history of testing for the audibility of audio signal polarity is littered with examples of strong adherents citing results of experiments which prove or disprove the audibility of it. In every case Ive ever seen, these experiments are conducted by reversing the connections somewhere in the audio signal chain -- in the so-called analog domain, which is the normal operating mode for most audio systems outside the guts of CD or DVD players and DACs. Examples of these analog-domain signal inversions would be: reversing connections at the loudspeakers; reversing connections at the amplifier; using a switching box with well-designed contacts for noiseless switching; switching the polarity of the connections at a phono cartridge; using a polarity-reversing circuit in a preamp (or black box) that is assumed to be transparent.
The problem is that all of those test methods are fundamentally flawed. They forget to account for things that do alter the sound. Because of this, none of those methods are adequate for evaluating the audibility of audio signal polarity reversal. Because none of these methods are acceptable test methods, none of the research done in the past using any of these test methods has achieved anything remotely close to proving whether you can hear polarity reversal or not. Any audiophile who has tested himself using any of these methods has also proved nothing about whether he personally can hear polarity reversal. What is it that researchers and audiophiles forgot about when listening for reversed polarity using any of these methods?
When you reverse connections at a loudspeaker, yes you reverse audio signal polarity. But you also reverse the direction the audio signal is applied to wire in the loudspeaker and to capacitors, inductors and resistors in the crossover. When you reverse wires, capacitors, resistors and inductors in the audio signal path, the result is an audible change in the sound. In the case of wire directionality, the difference can be just as large as the differences people ascribe to correct versus incorrect audio signal polarity. None of the testers/researchers who performed tests using analog-domain switching ever seem to have evaluated the audibility of the reversal of direction if signal travel in wires or electronic components as a separate issue from the audibility of audio signal polarity. All of them apparently made the same mistake -- assuming that audio signal polarity reversal is the cause of the change in sound they hear.
It is not difficult to evaluate the sound of reversed wire directionality. Just take any pair of audio interconnects and install them backwards, then have a listen. The difference in sound is not huge, but it isnt hard to hear in most audio systems, even systems based on mass-market electronic components like receivers.
Evaluating the sound of reversed capacitors, resistors and inductors is harder to evaluate than reversed wire directionality. You have to be a bit techie and interested enough to set up the experiment. I have done that and find the differences to, again, be audible and repeatable but also not huge. Capacitor manufacturers, loudspeaker manufacturers, and manufacturers of audio components have known for years that capacitors do not sound identical when their "direction" is reversed and these manufacturers that have taken the time to determine which direction sounds best in their product(s) will actually specify the orientation of the capacitor in their assembly procedures. Some manufacturers will evaluate wire directionality by connecting an entire spool of uncut wire to an audio signal and listening to the signal in both directions then marking the spool with an arrow indicating the direction the wire on that spool should be used in during assembly. A few manufacturers even mix wire directionality in their circuits to fine-tune the sound they get. This is a body of knowledge that cannot be dismissed when evaluating the audibility of polarity reversal.
Any time you try to evaluate the audibility of audio signal polarity reversal, you must use a test method that changes nothing other than audio signal polarity. This is virtually impossible in the analog domain. The problem in the analog domain is that any and every change you make to the path the signal travels in alters the sound a little bit. This fact may mean that polarity reversal in the analog signal domain, no matter how well intentioned or designed, is always going to contribute something to the sound you hear. Because of this, using any audio-signal-polarity-reversal method that changes the polarity in the analog domain must be dismissed from the list of valid tests until it is verified as sounding identical to only valid test methodology for audibility of audio signal polarity reversal -- digital domain polarity reversal. Further, anyone who believes that their analog domain polarity switching method is 100% transparent must test it against the only valid test methodology -- digital domain polarity reversal. Only if the analog domain switching method is identical in sound to switching polarity of the audio signal in the digital domain can that analog domain switching method become an "acceptable" test that reveals whether audio signal polarity reversal is audible or not. Simply designing an ideal analog domain switch and saying it is 100% transparent is not good enough. It must be evaluated against digital-domain polarity switching to determine whether the analog switch really is an acceptable method with no sound of its own.
Digital polarity reversal The gold standard in evaluation of audio signal polarity reversal
When you reverse audio signal polarity in the digital domain, nothing else in the audio signal path changes. Nothing changes but the audio signal polarity. To do this at home requires a CD player or DAC that has a polarity reversal switch on it. If polarity can be reversed using the remote control, so much the better since you can sit in your chair and reverse away to your hearts content. People who evaluate the audibility of polarity swapping in this manner, especially those with remote-control swap-ability, give you a very different answer from those who use swapping speaker cables or other analog-domain switching. The digital swappers find incredibly elusive sonic differences, some even admitting to not being able to hear anything different at all when polarity is swapped digitally. Some of these digital swappers were originally polarity believers when they swapped speaker cables and heard significant differences. Why do they hear little or nothing when swapping audio signal polarity in the digital domain? Could it be because for the first time they are really only hearing polarity reversal and not the sum of a bunch of wire-directionality changes and directionality changes in electronic components? I believe that this case is compelling. It does not give 100% proof that audio signal polarity is 100% inaudible. But it certainly does prove that audio signal polarity audibility has nothing to do with what you hear when you reverse speaker cables or phono cartridge leads or use polarity switches in preamps.
More information you may not know How do we hear?
People who have been promoting the audibility of audio signal polarity have been assuming that the hearing mechanism is a continuous analog process. Under this set of assumptions, compressions and rarefactions have equal weight within the hearing process and the order in which they are received in the ear would be quite obvious.
Mike VansEvers encountered a book (referenced later) that has a brief section that indicates that our assumptions about how the hearing mechanism works is not accurate. The book indicates that when you place instrumentation on the auditory nerve that transmits sound from the ear to the brain, you can see the stimulation the ear receives from the compressions and rarefactions detected by the eardrum. The first thing you notice is that this electrical stimulation is not continuous. There are bursts of electrical signal followed by no signal repeating in step with the sound being heard. These nerve pulses are tied only to compressions or only to rarefactions, never both unless the audio frequency is approximately 5kHz or higher. These nerve impulses are generated only on compressions or only on rarefactions. Furthermore, the ear "locks" randomly on compressions or rarefactions. The ear does not limit itself to only compressions or rarefactions. It picks one or the other randomly and switches back and forth. But it only responds to one or the other unless the audio frequency is higher than approximately 5kHz. This book does not specify how often the ear might swap ends, so to speak. So in spite of our assumptions that the ear hears both compressions and rarefactions, there is scientific data that proves ears do not generate continuous nerve impulses from approximately 5kHz and lower. Above this approximate threshold, the ear does indeed produce continuous nerve impulses which track compressions and rarefactions. There isnt much music above 5kHz, so virtually all instrument fundamentals fall into the range where our ears are firing only about 50% of the time on either compressions or rarefactions but never both.
The book where Mike VansEvers found this information is titled An Introduction to the Physiology of Hearing (James O. Pickles, Academic Press, sixth printing 1999). Mr. Pickles is in the Department of Physiology, University of Birmingham, Birmingham, England. The section revealing this aspect of the physiology of hearing lasts for only five or six pages and there is no other information on the audibility of polarity than what Ive encountered in these few pages. So if you seek out this book in hopes of hundreds of pages of revelations about this subject, you wont get it. But you do get this one chunk of scientific information about how the ear really functions. This information would seem to imply that switching polarity could very well be inaudible, though there might be some possibility that polarity could be audible above the threshold if the information in the recording was sufficiently rich in content.
The forces for the audibility of polarity claim that bass is one of the areas most easy to hear polarity differences. Bass also happens to be the zone of hearing most clearly not affected by changes in compressions or rarefactions. Im not saying that people claim that polarity is audible are hearing things. They are definitely hearing what they say they are hearing. The problem lies in their assumptions. It seems likely that what they are hearing is not what they think it is. It isnt polarity, its all those changes in wire directionality and direction of electronic components.
Proving the point
If you reverse audio signal polarity two times, thats the same as not reversing it at all. So you can easily devise an experiment that will prove whether reversing speaker cables reveals the sound of audio signal polarity reversal or something else.
In a system you are very familiar with the sound of, reverse the speaker cables and switch the polarity in the digital domain at the same time. You have then reversed polarity two times. This means that there is no polarity reversal. Do you still hear a difference in sound? If so, and I suspect that everyone who tries this will hear a difference, then you cant possibly be hearing the sound of audio signal polarity reversal. What this experiment proves is that wire directionality and electronic-component orientation is audible. It still doesnt prove 100% that audio signal polarity reversal is inaudible. You must still perform the digital domain polarity reversing experiment to know whether audio signal polarity reversal is audible or not.
Can you hear me?
How can you expect to hear the reversal of audio signal polarity if your loudspeaker is designed in such a way that half the drivers are connected with normal polarity and the other half are connected with reversed electrical polarity? In a three-way loudspeaker, the midrange driver polarity is often reversed compared to the tweeter and woofer, which will match. This practice is common in and out of high-end audio. How can you expect to hear polarity reversal if the reference recording is mixed up itself, as many commercial recordings apparently are.
To control your experiments, you really need to use recordings that are known to be polarity correct. This will generally require a simple setup: perhaps two microphones and no mixing board. Youll also need loudspeakers that have all drivers connected with the same electrical polarity (not all that common, unfortunately) and preferably loudspeakers with little or no phase error since polarity reversal and 180-degree phase shift produce identical results electronically.
I dont hear digital-domain polarity reversal, so can I forget about polarity then?
No! No way! There is still the issue about playing music as it was intended to be played back. There is a symmetry and rightness to establishing and maintaining proper system electrical polarity. Maybe it only makes a difference 5% of the time (a complete guess) -- thats enough to justify maintaining correct audio signal polarity in your audio system. I dont have a convincing overriding argument for maintaining proper audio signal polarity. It just seems like the right thing to do.
Some attentive readers may have noticed that if polarity reversal is not audible, perhaps all those loudspeakers being made with alternating driver polarity really arent a problem after all. Or maybe they still are. In the crossover zone you would have one driver contributing rarefactions while the next driver would be contributing compressions. How confusing would that be to your ears? It isnt going to jump out at you for sure or nobody would be alternating driver polarity. If it is audible, it is without doubt at some lower, perhaps long-term, level.
The intention here is not to advocate that audio signal polarity is inaudible or audible, but to get people on the right track when discussing the issue and to give reasons why you really can only rely on digital domain audio-signal-polarity switching to learn the truth. In my personal experience, Ive had very bad luck at hearing any difference at all when switching audio signal polarity in the digital domain. I used a modified CD player with remote-control polarity switching. I confirmed that the CD player did indeed reverse audio signal polarity using an oscilloscope connected to the outputs of the DAC and observing the waveform of the tone bursts from a test CD. If there is something going on when polarity is swapped, it sure isnt happening down where most of the music exists.
Copyright © 1999 SoundStage!
All Rights Reserved