subwoofer feedback system

[mazurek]

Been working for a while on a subwoofer feedback system, matlab makes finding the transfer function easy. I'm also amazed using the matlab tools that anyone can get a stable system without carefully computing the transfer function.

I'd like comment on a motion sensing system. It involves shining an IR led at a white to black gradient glued to the subwoofer cone, normal to the surface. After testing, I could modify the gradient pattern to account for moderate non-linearities in the sensor. A better variation would be to use a 50% reflective mirror, then the beam reflecting off of the subwoofer cone and coming back through the mirror and hitting an optical sensor.

Look at figure 7 on the optical reciever datasheet for the inspiration. http://www.digikey.com/scripts/DkSea...531880&Site=US
If the digikey link does not show, look for OPT101 in their search.

I could use these:
http://www.edmundoptics.com/onlineca...=2035&search=1

Alternately, if this idea fails, I have also looked at mems and piezo accelerometers. In my opinion though, their bandwidth is disappointing. I'd like to push the inductive/directional rolloff of the woofer further off to eliminate phase effects at a 100-200hz crossover point. Using analog devices accelerometers, I could push the rolloff to 2000 Hz, but these devices have issues with max accelleration and would have to be mounted off axis to measure the actual max acceleration, and still might not have enough range. I looked at measurement specialties inc accelerometers, these have good max acceleration, and high end bandwidth, but have some phase effects due to their 2Hz highpass filter.

[bobo1on1]

A really cheap solution is to glue a condensor microphone sideways on the cone, it works really well, sensitivity is good, bandwidth is good, but I read here that you should be carefull when closing doors.

The biggest problem I have been facing is that a woofer acts like a second order highpass that has a 180 degree phase shift at a low frequency (usually around 15 hertz), I have been trying to think of a way to equalize this but I didn't come up with a good solution.

[mazurek]

I was thinking about a microphone possibly, I was adding up the cost to do a led reflector system, and to do it real well will cost ~150 dollars per servo controller. Thats a bit much, and if I want more than one sub/woofer per speaker, that is getting to be quite a lot. I'll search for condenser mikes on digikey. Thanks.

You shouldn't need to equalize the phase shift to be able to produce a stable loop. I've just been solving the feedback transfer function, and plugging it into matlab 'sisotool' to tell me whether or not it is stable. The transfer function for the system is (Speaker*Controller) / (1+Speaker*Controller) I believe. Then you just solve for the controller you need to get the desired transfer function. The stable controller results I had usually involved a linkwitz transform type circuit to move the low end rolloff where I wanted it to be acoustically, and a lead/lag type controller to move the inductive low pass rolloff out, but it has to be lower than the bandwidth of the driver or it results in instability. I also found that the lower I wanted the system f0 to be, the lower I had to set q for the system to remain stable.

If you don't have matlab, I highly recommend getting a copy in order to easily analyze the transfer function. If you do have matlab, I can send you the script I was using to find the controller function.