1. Limited time only! Sign up for a free 30min personal tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Help Needed: Simple Audio Cancellation

  1. Dec 2, 2007 #1
    Some background information:
    I am trying to cancel audio signal/noise in an enclosure I have constructed.
    I have two identical woofers positioned opposite each other in the enclosure.
    I have a very accurate sound level measurement microphone positioned between the two drivers.
    I am using sound measurement software to initiate a sweep of the frequency response range of the woofers.
    After several trials, the graphs of the speakers (unadjusted) proved to be almost perfectly the same, which leads me to believe that me hardware is not lacking in quality.
    I then proceeded to switch the banana clip inputs from one of the speakers on the receiver/amp from red>red and black>black to red>black and black>red. I left the other speaker alone, and its inputs maintained their designed locations.
    Next, I ran the same audio sweep tests and compared the graphs of the one speaker reversed tests to those of the "control" all speakers normal tests.
    Oddly, the graphs of the speaker reversed test were rarely of less amplitude then the normal tests. Only in some places on the graph, was the reversed speaker test lower than the normal test, as I thought it should be.
    In most cases on the graph, the amplitudes of the normal test were less then the reversed test
    Correct me if I am wrong, but I thought that if I reversed the terminals on one speaker, and therefore changing the phase 180 degrees, it would yield considerably less noise in comparison with both speakers operating in phase.

    My main goal here, is to cancel noise.
    Can someone correct my theory and point me in the right direction?

    Many thanks,
    Last edited: Dec 2, 2007
  2. jcsd
  3. Dec 3, 2007 #2


    User Avatar

    I think you could use an op-amp to combine an opposing noise signal with the actual signal and remove the noise, but don't quote me on that.
  4. Dec 3, 2007 #3
    Unfortunately, simply placing speakers out of phase will induce destructive and constructive interference at various places, and thus simply cancel or boost both noise and signal equally at those places and each frequency.

    The only way to remove noise from the original signal is by creating a circuit or mechanical device that can somehow recognize the difference between signal and noise and cancel it--for example with out-of-band noise (like hiss going in to a woofer) you might simply use a frequency domain filter. In-band noise (in the original signal) cancellation requires a circuit or device intelligent enough to somehow recognize the difference between signal and noise, if possible. For example, if the "noise" were really simply distortion, then a special circuit or device can be designed to undistort the signal as long as the original distortion doesn't cause information to be lost. But by "noise" I'm assuming you mean unwanted in-band energy in the signal which is uncorrelated with the signal (distortion is correlated with the signal, so its easier to recognize and thus remove).

    There are two reasons you might see frequency response change when you swap the phase to one speaker:

    1. If you run your response test while driving both speakers with similar (stereo) or identical (mono, or one stereo channel) signals, then note that the phase you choose for the speakers will also alter the impedance of each speaker, and at different frequencies (the cone moves less or more, depending on the wave from the other speaker--that translates to a change in how much power the speaker converts to sound). So you'll see a different response depending on the phase you choose. Also note that a circuit designed to flatten the response of a speaker in an enclosure will be different depending on the phase of the speakers (assuming at least some correlation between the signals applied to each speaker).

    2. Speakers are non-linear. If the signal you used to measure response was not symmetrical, then the distortion from that non-linearity could affect the response. The non-linearity becomes more noticeable with the power. Try your test with very low power (and don't drive the other speaker at the same time) and I believe you'll see negligible correlation between the phase you choose and the response for a given single speaker.
    Last edited: Dec 3, 2007
  5. Dec 3, 2007 #4


    User Avatar
    Science Advisor

    if I read you correctly, you are trying to cancell airborne soundwaves with other airborne soundwaves (as opposed to cancelling signal before it reaches the speakers), right? This is next to impossible for the setup you have. The speakers are sending out sound waves in all directions at once. If you have the right frequency and amplitude to cancel out a sound wave in one location, you will find that you've amplified that soundwave almost everywhere else in the room. Have you tried moving your microphone around the room? You may discover that there are certain "sweet spots" in which you are successfuly dampening sound. But maybe they won't be obvious or even measurable, since sound is bouncing all over the inside of the room.
  6. Dec 3, 2007 #5
    Is this some piece of industrial equipment generating unwanted noise?

    You will get more results out of insulating foam to dampen the vibrations and sounds then you will out of a noise-canceling circuit, although using both together is best.
  7. Dec 4, 2007 #6
    Back in the 1970s, I saw a similar scheme tried by an aluminum slitting plant (an extremely irritating, damaging noise). The problems they kept running into were: (1) it was never possible to put the cancellation speakers at the exact location as the noise generators and they had interference and reinforcement patterns that drove people half crazy (and somewhat deaf); and the slightest drift in the electronics (much less of a problem today) generated two godawful sounds instead of one.

    Damping ended up working much better.
  8. Dec 4, 2007 #7


    User Avatar
    Science Advisor

    Probably the only way to cancel soundwaves for everyone everywhere in the room is to have them where sound-cancelling headsets. That's the only way to make sure that the area of maximum noise cancelation coincides with the location of an eardrum.
  9. Dec 5, 2007 #8
    Thank you all for your suggestions.
    Here is one big answer to all of the responses:

    Yes, we have considered using electrical apparatus involving opamps, resistors and pentiometers, to alter the cancellation of certain frequencies, but we need some guidance in the area.

    Yes, thank you.
    You have provided much useful information.
    I was aware that the speakers, even if out of phase would cause destructive and constructive interference in different places, depending on where the person is listening from, and the frequencies being played. However, in this case, we would like to only work with about three frequencies (which can be considered low frequency noise) and have the microphone stay static, so the listening place would stay in the same place. Thusly, we would only have to work with canceling audio for that one place and at several frequencies.

    Thank you.
    I realize that using my setup in a room and trying to cancel sound waves actively is not feasible. As I said before, the testing apparatus is a small enclosure, with two woofers facing each other and a microphone between them. (Pictures are attached)

    Yes, I understand, but we want to pursue active noise cancelation, not passive.
    No, there is no nagging noise, this is an experiment I am doing. I will be playing a steady frequency through both woofers (Mono).

    Yes, in a lot of industrial situations I would tend to agree.
    However, we are not trying to cancel a nagging noise like an HVAC system, just pure tones.

    Thank you for the post.
    That would make sense for in a room, but as I have stated, we are working with a small enclosure, and only one listening point.

    To clarify, I have watered down our actual experiment and procedure a fair amount, I am just trying to figure out a way to get this active noise cancellation system working.
    This is just a proof of concept device. Pictures of the enclosure are attached.

    Again, any help would be greatly appreciated.


    Attached Files:

    Last edited: Dec 5, 2007
  10. Dec 5, 2007 #9
    Hi Sam,

    Unfortunately (I've got to stop starting my posts with that word), there are two popular definitions for the word "noise" and it looks like that ambiguity has bitten us here.

    A popular definition can often be "unwanted information that is easily predictable or detectable separately from any signal". This is the definition used in the phrase "noise-canceling headphones". The noise is easy to cancel because it is separate from the signal--which in the case of the headphones would be silence, music, what have you.

    In information theory and electronics the definition is closer to "unwanted unpredictable information completely mixed with wanted unpredictable information, where there is no trivial way to recognize the difference in the two".

    There is certainly a similarity in those definitions, but the differences in the definitions are critical. The first definition implies the noise can easily (at least theoretically) be recognized and thus canceled. The second implies it cannot. I'm now thinking you meant the first definition?

    Those images have not yet been approved and I'm not familiar enough with the PF process to know how quickly they might be. I also see that there might be some proprietary issues for you, and of course that does make things hard. But if you get a chance and feel comfortable with revealing the information, can you answer a few questions (answer only those you feel comfortable with)?:

    1. What generally is the source and nature of what you are calling noise? Do you have any, or even complete, knowledge of its pattern? (thus making removal a cinch)
    2. Is there some signal information you are trying to separate from noise? If so, how much can you predict its pattern (how redundant or compressed is it)?
    3. If there is signal, how mixed is it with the noise? For example, if you used two microphones cleverly placed or pointed you might obtain two channels with the same signal and same noise but different SNR, then you could separate the signal and noise easily.
    4. Are we talking about something like the way noise-canceling headphones work?

    And one comment that may or may not apply:

    The human brain, itself, has (so far) been able to separate signal and noise (using the second definition above) far better than any device the human brain has created. So if the goal here is stealth, I doubt a device can win against the human brain unless the device knows something about the phenomenology of the particular noise and/or signal that the competing brain does not. For example, in order to obfuscate signal in a channel, a device might inject, or modulate with, noise and then remove that noise using its knowledge of the otherwise secret phenomenology of the noise.
  11. Dec 6, 2007 #10


    User Avatar
    Science Advisor

    I think the "noise" he's talking abotu is the sound from the speakers. That makes it very predictable, as it is being generated for the purpose of the experiment. Ocacoustics, you are thinking that, because both speakers are producing the same sound wave, at the same time, in opposite directions, then there should be a "quite spot" (or at least qiet-er) exactly mid-way between them, right?

    But I think the mere fact that soundwaves are bouncing off the walls inside your enclosure, and arriving at the mic out-of-phase from the waves coming directly to the mic, is going to make things impossible for you. The same waves that cancel some incoming soundwaves will end up reinforcing others.

    Just out of curiosity, have you tried moving the mic around, and looking for variation in sound levels? Could be some useful data obtained that way; maybe you are getting partial cancellation, but just not in the exact spot you expected it.

    Also, thinking about the reflection problem mentioned above, have you tried this without an enclosure?
  12. Dec 6, 2007 #11


    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    If you're only interested in cancelling sound in one spot, and only at some specific frequencies, I suggest you do the following:

    Drive both speakers with the same signal, but one delayed with respect to the other (digitally, perhaps). As you vary the amount of delay, you will eventually find a point at which the two signals are 180 degrees out of phase at the mic, and cancel.

    Note that this will only work, in general, for pure sine wave tones.

    - Warren
  13. Dec 6, 2007 #12


    User Avatar

    you can also reverse the polarity (swap the leads) of one of the speakers and you should be cancellation at every point that is equi-distant from the two.

    doesn't always work.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook