Laser Instead of an Accelerometer? Is it possible ?

[jsa_ind]

Is it technically possible to develop a servo subwoofer using a laser system such as http://www.opti-cal.com/heads.html rather than a conventional accelerometer ?

If so what would the principle be like ?

All advise would be greatly appreciated !

Thanks !

Junia.

[mazurek]

I have thought of such a system. It would use an infrared laser pointed at a 50% reflective mirror pointed at a 45 degree angle. The beam would bounce off of the woofer cone, and back through the the mirror into the reciever under the mirror. The beam would aim perpendicular to the woofer, and the woofer would have a white->black gradient printed on it. Parts would cost about 100 dollars for everything.

[soongsc]

Technically it is possible. If you can get it to mount in the center of the motor and measure the dustcap distance, the design would be clean.

[rpapps]

Originally posted by mazurek
I have thought of such a system. It would use an infrared laser pointed at a 50% reflective mirror pointed at a 45 degree angle. The beam would bounce off of the woofer cone, and back through the the mirror into the reciever under the mirror. The beam would aim perpendicular to the woofer, and the woofer would have a white->black gradient printed on it. Parts would cost about 100 dollars for everything.

Hi
You've got my interest.
Care to share?

[Svante]

I think it is a bad idea. The sound output of the speaker is not proportional to the position of the cone, but rather its acceleration. So, if the position of the cone is detected, a double differentiation would be required in the feedback loop.

It appears much smarter to me to detect the acceleration of the cone in the first place.

(Nevertheless there was an article in JASA recently where someone had made something similar, detecting cone position)

[mazurek]

I cannot find a stable transfer function using acceleration as feedback, with a desired target function. The only unconditionally stable transfer function I get is when the feedback is position, if I were to use an accelerometer I would double integrate it.

If I were using an accelerometer, I would use this:
Digikey part number: MSP1001-ND

From the specs sheets of the analog devices imems devices some recommend, I believe that they have just barely enough measurement range, and the bandwidth is a little disappointing (i.e. only appropriate for subwoofer correction and equalization below one-two hundred hertz IMO). I believe that they could be used though.

If you have matlab, please look over the scripts I built to test variations of the feedback system. It takes a given target transfer function, and finds the resultant controller, taking into account the feedback transfer function and the speaker transfer function. I then use the matlab 'sisotool' which is a feedback system design tool, it reports on the stability of the resulting system, and can be used to experimentally move poles and zeros and explore the response.

I am somewhat concerned about the results some people post on feedback systems. To me, it appears on this forum that most do not take into account that the closed loop transfer function

with controller in forward path
= (speaker*controller) / (1 + speaker*controller*feedback)
with controller in feedback path
= (speaker) / (1+ speaker*controller*feedback)

So for those that just tack on feedback into their system, and just adjust the gain until its stable, they are completely changing the system transfer function.

[jackinnj]

Originally posted by Svante
I think it is a bad idea. The sound output of the speaker is not proportional to the position of the cone, but rather its acceleration. So, if the position of the cone is detected, a double differentiation would be required in the feedback loop.

It appears much smarter to me to detect the acceleration of the cone in the first place.

(Nevertheless there was an article in JASA recently where someone had made something similar, detecting cone position)

A laser has no "inertia" to speak of, at least as contrast with an accelerometer.

[mazurek]

These are the parts I would use to build a laser system:

Half reflective mirror:
http://www.edmundoptics.com/onlineca...=2035&search=1

Reciever:
Digikey search: OPT101

I haven't picked a laser yet, there are some good options in the range from 10-30 dollars
Search at digikey under "laser diode" or "infrared laser diode".

The color of the laser probably doesn't matter.

Edmun optics also sells mounting brackets for their mirrors if something could not be fabricated adequately.

[mazurek]

Here is a picture of the test patterns I had printed on matte paper.

I do not believe printer resolution is an issue due to bleeding which averages the value, and the beam size and pickup area which also averages the value. There are two styles depending on the ultimate setup I choose to use.

One advantage of this optical system is that the picture gradient can be modified to account for linear errors in the transducer system, and the gradient can also be modified to avoid subwoofer destruction. For example, different shades can be printed near xmax to eliminate destructive feedback, or to apply feedback to keep excursion under control.

[mazurek]

Here is a picture of my original concept. Not sure if this would fit in the motor to measure the dustcap. I had envisioned just using the cone as feedback. I feel this would work under the cone breakup mode, and possibly over the breakup if mounted in the correct place. As drawn, it depends on having a linear cone cross-section, however the gradient could be printed with a compensation function for a curved cone.

One advantage of this system over others is that there are no cords whatsoever hanging from the cone, and quite insignificant weight. Additionally, it can be done fairly cleanly with only analog processing (not necessarily a sound advantage, but developing a digital platform takes even more work).

The reason I have not yet gone further with this is that I am redeveloping my amplifiers/crossover to package them together in a more conveniant testing platform. Also, my subwoofers do not have a removeable baffle, therefore it is quite a pain to try this out. I was thinking about getting an IR beam and filter to maybe mount this on the front of the woofer to test. This project was on the back burner until I get my amplifiers changed. If someone wants to co-develop this stuff, it could be expedited.

I forgot to say, the feedback controller function ends up being in the form of two lead lag controller and a linkwitz transform, not to say that it is an actual linkwitz transform, just that is what the transfer function looks like.