Could a pair of headphones be deisgned like this to completely eliminate harmonic response... - Page 2

 

You won't fully eliminate it anyway, and while .1 is greater (I think more are around .5-1%), I doubt it would be audible.  That's headphones and not speakers though.

How would you keep all those drivers in phase? And I suspect that the crossover would be ridiculously complex and start generating problems in itself.

See, that's the beauty.  You have special 22-channel stereo recordings commissioned that have a discrete channel for each frequency range on each side.  No crossover, and since the design would be active, each DAC and amplifier could be specifically designed to perform in its frequency range without rolloff until outside the range that the signal contains.

 

Nevermind the addition "fun" that such a setup would entail!

Don't take it too hard. You are thinking along the right lines and asking questions. Keep doing that.

This is just the nature of the beast in audio. You can optimize one thing and other engineering problems crop up. If you have too much complexity, it starts to bog down.

If you haven't poked around with crossover designs and phasing, there's a lot of interesting reading out there. For crossovers, check in at Linkwitz Lab. That is one guy who really knows his stuff. You'll learn a lot and it might convince you to build a Pluto or Orion.

 

 

Yes, that would be the cause of loudspeaker-sourced harmonic distortion, and you would have to get rid of all the resonances and other causes of distortion (I don't know them myself) to eliminate them - quite simply, that's not physically possible.  You could minimize it, but everything in a loudspeaker is a compromise - you can never eliminate it.

 

Nor can such a design ever eliminate THD created by the electrical components.

 

 

 

However, I have found an interesting paper on the use of DSP in reducing the harmonic distortion of a system.  I don't really have the time or desire to read through the whole paper, but my understanding of what they're doing is to measure the harmonic distortion of a system and then compensate for it in processing by introducing opposite phase waves to the expected distortion.

http://www.nokia.com/library/files/docs/Simplified_Loudspeaker_Distortion_Compensation_by_DSP.pdf

 

 

 

Also, if you're interested in research into human perceptibility of nonlinear distortions, you may be interested in the research that Drs. Earl Geddes and Lidya Lee have done:

www.gedlee.com

 

 

 

 

 

I'm just trying to highlight the complexities inherent in such a design, and as UE has pointed out, the very obvious potential for the problems created to outweigh the problems solved.

 

Here's a good site that explains the problems inherent in crossovers:

http://sound.westhost.com/ptd.htm

 

Honestly, I hardly know anything about active crossovers and how they compare in practice.  However, they do still impact the phase of the signal, and I don't even want to start to think about how complicated and problematic such a system would be.

 

However, in the digital domain you could get around the problem, and that was the legitimate point I was trying to make through my sarcasm.  You'd still have to deal with the distortions induced by the electrical components, but you would be able to bypass the issue of the crossover.  There's a tad bit more complexity to the system, however...

 

Oh, and I forgot about this:  You'll have serious problems with imaging once you start having that many drivers, especially with the distances between them and the overall size of the loudspeaker.  There may be some way around this, with a multi-horn design or the canal in a balanced armature design - but again, we're introducing complexities.

 

 

 

 

Now do you understand why some people are fans of single driver speakers and full range electrostatics?  They certainly have their own problems, but my point is that every design is a compromise.

There are lots of interesting designs out there that try to overcome the resonance and other issues inherent to speaker cabinets. I can't find it at the moment, but came across a fascinating design once that used a bunch of tuned traps to control the problem. Like a dozen or two of them. I wish I could find something on this. Apparently, a few people (through painstaking trial and measurement) managed to eliminate the resonances. Apparently, it works well, but requires so many calculations and fuss that it's difficult to design. Further, the cabinet work was very complex, making manufacturing difficult.

I spent a lot of time deadening the cabinets on the ProAc Respone 2.5 clones I built. I lined the interiors with Dynamat then glued lineolum floor tiles over it before adding stuffing. At each layer, you could knock on the side of the cabinet and hear the pitch change. Even the amount of stuffing used changed the sound. Not the most complex design, but very educational in showing how even slight changes make a difference. That was a rewarding build.

Another approach I like is Linkwitz' approach to the Phoenix and Orion. They're open-baffle, which eliminates a lot of the resonance problem. Of course, dipoles raise a bunch of other issues. They're fiddly with placement and you have to carefully plan the cabinets so the sound doesn't cancel itself out, and much else.

Ethan Winer is the guy you should ask about this. He's on Head-fi, but doesn't do PMs apparently. You might drop in on a Sound Science thread he's posted on and direct him over here.

You'd create more problems than you'd solve. Passive x-overs would be far less efficient and would introduce phase problems; active and/or DSP x-overs would ramp up the cost significantly.

Ramping the cost significantly is kind of the point. :P  The only problem that this is trying to solve is the THD, nothing more and nothing less, really...I doubt this would be viable for any practical usage anyway, other than having a perfectly balanced set of speakers that cost $1,000,000+ for parts alone. :P

See... if you ping a speaker, and it has a penchant to ring, it will. Even if it is outside of the signal you're feeding it. So, you're idea of getting rid of harmonic distortion by limiting the bandwidth of the applied signal sort of misses the point.