Would ANC Work Better With 2 Separate Drivers?

Headphones with Active Noise Cancellation (ANC) circuits are often shunned on Head-Fi, because the technology that makes the noise cancellation possible also has a tendency to interfere with accurate sound reproduction, and from my experience, can make them sound very sibilant at times. However, having a good ANC can be undeniably useful, especially in loud environments such as airplanes or mass transit, where the outside noise would interfere excessively with the music, and for people who are uncomfortable with using IEMs to block out noise. 

This brings me to my main point: Would a headphone with ANC work better if it were equipped with 2 separate drivers in each ear, one for the ANC, and one for the audio?

Here's my rationale: It seems like a big reason why the ANC doesn't work as well is that the driver has trouble accurately producing the sounds required for the ANC and the music from your audio source at the same time, which results in that non-audiophile distortion and such. This seems to be a physical limitation of the driver, as the signals required for the ANC add to the complexity of the movement of the driver, which overwhelms it and causes it to lose detail. 

This problem could be remedied with a faster driver, but that might be difficult to do. Instead, my idea is to have a dedicated ANC driver, which could eliminate this issue by delegating the 2 different signals to two different drivers, reducing the speed requirement of each driver significantly. That way, the driver reproducing your music would not be burdened by the extra ANC signal, and could play music to its full potential.

Also, because the ANC and music circuits would be completely separate, you wouldn't need any complicated crossovers, and you could always turn one off without affecting the other, an added functional bonus.

So, what do you think? Would this work, or am I barking up the wrong tree? Or has this already been done, and I'm just ignorant?

P.S. I'm adding the poll just for kicks, and so it's easier to gauge the general consensus on the issue.

Bump, anyone? Also, if you vote, can you just give a quick reason why? Thanks.

Quote:

Originally Posted by anetode 

ANC is preformed via a mic & DSP. It's simpler to add the compensation to the original signal. The prospect of simulating the interaction of two different drivers in a single enclosure is pretty harrowing, especially if you're talking about running them over the same frequency range. Any variations in fit and the shape of the user's ear will alter the required signals, thereby causing all sorts of unpredictable destructive and constructive interference.

Why would it be more complicated though? I'm no expert, but I thought sound waves are additive, so having 2 drivers playing different sounds is the audible equivalent of one driver playing the digitally added sum of the 2 waves. Also, wouldn't the issue with ear shape also affect conventional ANCs?

Essentially, there would be 3 different sound sources in the cup. The ANC and the outside noise would ideally cancel each other out (assuming you have a good DSP, which I'll assume is a given), just leaving the desired audio source. I don't see the problem with that, but is there something I'm overlooking?

Quote:

Originally Posted by anetode 

Yes. Read up on closed-loop transfer functions, quantitative feedback theory and head-related transfer functions. Then purchase the following article. The math and amount of trial and error in designing a proper ANC solution are harrowing. I can digest bits and pieces of the science behind it, enough to know that there are no simple solutions to this problem and that it would take years of intensive study to get a good enough grasp to attempt to innovate in the field.

Your point on the inherent complexity of ANC development is well taken, but given an n already functioning ANC system, why couldn't you just add another driver to produce sound in that ideally zero-noise environment. I read the abstract of that article, and though it does give some interesting tidbits on pure ANC systems, they don't really discuss adding in other sounds on top of the ANC system. So again, my question is, how would the additional signal interfere with the ANC if the ANC and outside noise would theoretically have already cancelled each other out? I.e., if you took a pair of Bose ANC headphones and only used it for the ANC function, then put in a separate set of drivers (that we hypothetically assume don't interfere with the acoustics of the cup, for now), what would the interference be?

Quote:

Originally Posted by anetode 

That's the trick: Adding another driver would certainly change the system, however it would not necessarily improve it. Your assumption that adding another driver would produce favorable results would need to be examined on a theoretical basis and then tested in the real world. I don't know what the results would be.

 

It would be very difficult to model the possible interferences in the planning phase. Of course you'd start with a computer model to facilitate dynamical system analysis. You would then have to create a few prototypes, measure them in several different model environments, then round up a set of volunteers to gather a usable set of subjective data. If the subjective data does not concur with the pre-measured quantities, or if you collect some other type of surprising data, then the approach might be deemed a failure.

 

So basically what I'm saying is that I don't know if another driver would bring about an improvement. I can think of many reasons why it wouldn't work and hardly any justification for it being an improvement.

 

(For what can go wrong, simply visualize the space between the drivers and the ear surface as a liquid. The drivers are pulsating, pushing waves that bounce off each other, the headphone casing material and the ear. At the same time the ear is moving, slightly, the pressure on the earpads changes, as do their damping properties and all the while there is a background storm behind the headphone casing that's hitting on all sides. Also, you have to include the interference caused by the cup -- the cup is an integral part of the system. At this point in the simulation I hope you can see that adding more drivers would only cause a greater mess by tasking the DSP with more complex calculations to compensate and model the proper cancellation response.)

But other than the physical features of the second driver in the earcup, how exactly would it change the system? I mean, given that ANC seems to work to a certain extent in existing headphones, can't we just build off that research?

Quote:

Originally Posted by anetode 

Well yes, exactly: the physical features of the second driver would change the system. Maybe there is a way to effect a positive change, however that is far from guaranteed as this system you are describing is ill-defined and complicated. ANC isn't a magic fix, it is a multifaceted approach that uses a variety of complex algorithms. You're thinking about this backwards. If you happen upon a specific approach to implementing ANC which would benefit from a second driver, then you must have some solid theoretical basis beyond your assumption that it is the driver which limits the effectiveness of ANC. A well-designed headphone driver doesn't care about what kind of signal it receives, it will keep up within its physical performance limitations. I don't see how the mixed ANC signal would "overwhelm" it or why adding another driver with similar physical limitations would overcome the supposed problem. It's more likely that the problem is with the ANC algorithm itself, along with the limitations of the programming for the DSP and the limitations of the data picked up by a cheapo mic.

 

I've tried to offer a few trivial explanations based on my limited knowledge about how a second driver would add unneeded complexity (though a few headphones w/o ANC have utilized multiple drivers, even multiple driver types, along with a crossover to delegate portions of the frequency response to each driver -- again, not quite what you're talking about). Now I think that this premise is at least partly rooted in an overly simplistic conception of ANC where you're expecting a deus ex ANC to justify your proposal, when in reality headphone drivers are components of a greater system which must be considered. Would a second driver necessitate a second mic? Where would it be placed? If it's simply a driver which plays outside noise out of phase, how can you direct those pressure waves around the entire cup or the topography of the ear without encountering interference from the driver which is tasked to play the primary signal? How would it be powered?

 

You can probably find research on this very topic. If you do, it will likely be dense and to grok it would require a thorough knowledge of signal processing, psychoacoustics, etc.

 

From a cynical point of view I'd like to point out that this approach has not been embraced in the competitive ANC headphone marketplace.

Well, the original rationale for having the second driver was that the added data from the outside would make the signal too complex for the driver to handle, much in the way that complex music can sound muddy. The idea was by simplifying the signal going through the music-playing driver, you'd get better sound quality for the music. Of course, as you pointed out, this assumes that the dip in sound quality is caused by driver overload, which it well may not be.

However, I question if adding that extra driver would really cause such a detrimental effect on the ANC functionality. First of all, the dampened housing should eliminate a lot of the acoustic interference, and second of all, I don't think that the contours of the second driver would cause that much interference to begin with. Sure, the journals cite acoustic interference as a big problem, but remember, they're using 6-meter, undampened steel ducts, while we're talking about 10 cm, acoustically dampened headphone cups. 

Anyway, just to clarify, my original post was not so much about a fundamentally new way to do ANC, but just a different way to integrate it into headphones.

And the cynical side of me wants to point out that maybe nobody has implemented this because they figure people will buy their headphones anyway, and there's no point in spending money to develop better sound quality. Though they're probably right, since you tend not to notice these kinds of things in high-noise environments...

Quote:

Originally Posted by khaos974 

Try to see it this way:

 

- With a two drivers, you don't know how the soundwaves from the two drivers will interact with each other (because multiple reflections in an enclose space, possible frequency nodes), you also need to fit a second driver in a small enclosure. Both tasks are easier said than done.

- On the other hand, adding electrical signal is pretty straight forward, and single source is much easier to predict than dual source.

 

Also, the sound of your headphones don't suddenly sound worse when you play symphony compared to a string quartet, yet you are feeding 'more' signal to it. The same happens with the added signal of the ANC. The trick is to realize that the ANC or symphonic music are not feeding more signal to the driver, it is just feeding it a different signal.

Well, my premise was that the ANC waves would cancel perfectly with outside noise, so there's nothing to interfere with the second audio signal, so there shouldn't be an issue with sound reflection beyond what you find in a normal headphone (I'm assuming a perfect ANC here of course. But even if it's not perfect, the interference would be no more than the sound from a non-ANC pair interfering with ambient noise leaking into the cup). Also, though putting in a second driver is not as easy as I made it out to be, small portables like the Phiaton PS320s and the Skullcandy Uprocks prove that it can be done.

Also, think of it more like playing both a violin concerto and a movie dialogue over each other. If your driver isn't fast enough, it can't resolve the overlapping audio signals clearly, and the sound becomes muddled. Even if you can't hear the movie dialogue, the violins won't be produced to the full potential of the driver, because the driver has to produce both sounds at once. However, if you have one speaker playing the violin and a separate one playing the dialogue, if you ignore one the other still has maximum clarity. At the very least, the two separate speakers will sound clearer than the one trying to produce both.

At least that's how I'm conceptualizing it. Basically, the idea is that if each driver is dedicated to producing one noise, the system as a whole can do both more accurately.

Quote:

Originally Posted by khaos974 

Your interpretation is wrong, a drive doesn't reproduce signal A, then signal B, but both signals at the same time, that's how linear systems work.

Sorry if I was unclear, that's not what I meant. I just meant that the additive signal would be significantly more complex than either the ANC or music signals individually, and given limited driver speed, the additive wave would not be reproduced with the same accuracy that 2 of the same drivers could produce if they each played one. Just think about playing two songs through different speakers simultaneously v. playing them both through the same one.