RON430

Actually I was thinking of taking advantage of different drivers and delay times. No point in using an identical driver with a typical ANR setup in phase unless you are going for the sounds of silence. Percy's original post about putting the accelerometers on the woofer seemed to indicate that you would then use the signals to feedback and do something. It would seem that all you could do (simply) is alter the drive signal power across the board or (more complex) alter the drive signal power at specific frequencies. Have to think about what that would be trying to accomplish.

Haven't spent a lot of time looking at ANR or the DSP they have to use to drive it (Ok, so I have ANR headsets for use in the plane but I haven't felt like prying them apart) but the DSP is set up to rapidly produce the out of phase signal to cancel the unwanted noise after listening through a small pressure tranducer (microphone). If the point of this speculation is to find a way to get a driver to produce a cleaner signal, I just thought maybe a better (quicker-cheaper) starting place was the ANR technology. The AD accelerometers, if memory serves me, are surface micromachined comb drive accelerometers that AD also builds a lot of signal conditioning on chip. Actually working on a device that works in conjunction with those things right now for a microsatellite guidance system. They could do a measurement on parts of the driver, either cone or support or both, but then you need to do something with the measurement, like I mentioned before. As long as we are playin' around, I thought that the ANR setup might be a better starting point. Seems like you could play around with the ANR driven driver (maybe not even a woofer but a full range driver-or as close as you could get), specifications, location, etc. and maybe use the thing to dynamically clean up the awful auto environment (from a fidelity standpoint anyway).

Maybe somebody will go for it. Ain't going to be cheap or quick but it is great fun to speculate.

engin_ear

iTrader: (1)

Ron:
Ah, but you ARE going for the sounds of silence - the absolute silence of the error.

I think all that happens with the cone accelerometer servo is that you get a signal back from the accelerometer that is a scaled replica of the signal that was sent to the woofer. In the DSP sense, you then re-scale the signal and then subtract the received from the original, thereby generating an error signal, which is then negated and subsequently added to the woofer's input signal. This occurs across all of the woofer's input frequency spectrum. Don't need to worry about matching error phase here, since the freqs involved are low compared to the time required to generate the error signal.

Yeah, the drive signal power is altered, but instantaneously, so that what winds up happening is that if the error is occurring at specific frequencies, the error correction will occur only at those same frequencies, due to the speed of the correction signal.

As far as ANR goes, I've been doing quite a few ANR proposals here at work lately. The phase of what you generate for correction is absolutely critical in order to achieve proper cancellation within the volume of air in question, or at a particular surface. For the ANR headphones, the air volume is relatively small and the surface of concern is the eardrum, which is at a fixed distance from anything within the earmuff. This characteristic makes the job of cancellation more feasible. Also, the entire earmuff tends to vibrate equally due to the noise, since the source is not within the earmuff, and so this allows many places to effectively mount the secondary driver for correction. This is quite unlike a car, where you preferably need to cancel sound energy at many points, since you not only move the head, but there can be multiple passengers, and the sound source is within the car. So once you bring the "correction" driver away from the source of the noise (in our case, the woofer error), you then have to worry about phase across all points in space where you care about sound quality in the interior of the car.

It is much easier to correct the woofer's motion on the fly using its own voice coil than to let it flap and try to cancel it elsewhere.

RON430

Boy, isn't "the silence of the error" a wonderful way to put it. An ANR muff is a quite small volume but the technology scales up surprisingly well, with some disadvantages over the muff and some advantages, according to some acquaintances. NASA has an aircraft where the whole interior (think biz jet size) is a big ANR muff and they use drivers to cancel out noise from all the usual airplane sources. I have had a chance to be in the thing and it works surprisingly well. Actually a bit larger than typical car interior. But the engineers said it was a lot easier to put together than they thought.

Two issues being mixed, and I did it so maybe that wasn't a good idea. Making a speaker "accurately" reproduce a signal by itself is probably a more appropriate starting point. Back to the woofer topic, sensing vibrations in the support structure wouldn't be that difficult to do with a MEMs sensor, although it will likely pick up displacements that might not be due to the drive signal. Might be a bit trickier on the cone but the nice thing about speakers is that they are already magnetic sensors for their deflection, although I don't know if anyone has ever taken advantage of that. The most daunting aspect of making this thing work is the software for the feedback. Still think an ANR chip might be a close starting point as you are going to try and determine what the "true" response is, and then adjust the driver to cancel unwanted response out. The ANR gets paid by "listening" for a signal and amplitude and when it "hears" what it doesn't like (according to its code) it very quickly generates a counter signal. But ANR has a much easier task having to deal with a predetermined set of conditions qualified as "noise". How you do the code for the woofer to distinguish between wanted and unwanted movement with something other than test signals sounds pretty interesting. Still doesn't sound cheap or quick but maybe not too far out.

I know the results on the Vel but I am not sure I ever saw any description about exactly how it got it. With appropriate apologies, the Vel was put together a few yers back and you have to wonder if that driver box could be reduced to a smarter chip today.

Great fun this.

engin_ear

iTrader: (1)

Wouldn't I love to see (hear) how NASA does that...multiple cancellation drivers, I assume? Given this, an associated topic might be to cancel unwanted body resonances with a separate woofer. My bass sounds better when the car is still - I've attributed this to the body resonances, at speed, cancelling out some of the low bass. Whenever I stop for a red light, it sounds better. I'd love to cancel THAT out, and maybe some other environmental stuff too (tire noise, etc.) The relentless pursuit of sonic perfection, eh? I wish I had more time to spend on it....

Jerry.

RON430

Youbetcha. Not sure how many as they were pretty well hidden around the cabin. Can't remember if I was told any specifics but each seat had at least one driver and there were more in the walls.