Motional Feedback Speaker contraption and camera test 

And new Magnepan . Look forward to that .

There is a spike in the microphones frequency response at 2680Hz, makes the audio "whistle" a bit - for lack of a better word. its just a little too loud at that frequency. The Old camera did not have that. And yes, i used FFT Analysis on this video to pinpoint the frequency. You can ignore my comment, its just me nerding out on audio 😊

About the speaker - one of the things MFB does is to change the output phase behavior around the resonance frequency. Typically you have a 180deg phase shift from 1oct above to 1oct below the resonance, and this adds overall group delay. That group delay (or phase shift depending on how you see it) is what IMO changes the most with really low tuned, big sub boxes. Two 15" subs tuned to 17Hz have a much lower group delay (and phase shift of course) at 35Hz than a smaller sub tuned to 35 Hz.... So even if you play them both at exactly the same level, with content that contains NOTHING below 40Hz, you will hear a difference. At least I think it's pretty obvious... But it all depends on how the MFB is set up, and what the bandpass limitations of the feedback loop is. This is fiddly indeed, I've done this with industrial motion control motors positioning several hundred kg's with tens of thousands of Newtons of force. Then you REALLY get to know when you messed up :)

Excellent topic and video. I tried to do this ten years ago and was not able to get a very linear optical sensor. I was trying to use a line of light and then block part of it with a tiny shade that was attached to the cone. My attempt at a low noise preamp that included the double differentiator, required to convert the position measurement into acceleration needed for the feedback would of course amplify the high frequency noise from the optical sensor. It turns out individual photons hitting the sensor look like noise, so a bright light is useful to get a good SNR. In the end I decided it was easier to buy linear midrange drivers from SEAS that would produce 0.1% distortion without feedback. It will be interesting to see what kind of SNR, and distortion figures you can get. The measured wave on the scope looked to have a large non-linearity which i would expect if the amplitude is changing due to distance from the source. Very fun project. I also tried to use laser interferometry but the range of motion of a midrange or tweeter is huge. It covers the transition between less than one wavelength of visible light and multiple waves, and the waves can go by at rates above 50 MHz. So that gets tricky. Keeping the moving mirror parallel to the fixed mirrors would be difficult as well.

You have LED transmitters and receivers for the sensor device.

EQ sounds very different than motional feedback because EQ changes everything at every volume and MF only changes what error the speaker system has.

Large aperature lenses (small F numbers) reduce the depth of focus. It is usually better to get more light!

That G7 has a 1" sensor, which is about half the size of the APS-C in the Sony, so the f/1.8 G7 is almost exactly the same light input as f/3.5 on APS-C. A very easy way to explain why is that the ONLY thing that matter to the physics behind it is the size of the front pupil. Nothing you do behind the lens entry pupil can change the physics of light in front of the camera. So the Sony kit zoom is almost exactly the same in that regard, no better and no worse. And no, I didn't think the image looked "soft"... :) Congrats on the new camera In stead: Get more light - a lens with a bigger aperture will give you a very short depth of focus up close. That's really hard to work with in video unless you put a lot of work into it and really know what you're doing. Sony have some really good/reasonably cheap 35/1.8 for E-mount, but it's still 350e... Video and photography is often MORE about lighting something correctly than having the "best" and most expensive camera equipment :)

f 3.5 is fine for your use! ...OK, maybe focus is better with a f 2.0 ... but DOF @ 2.0 is really tiny for that kind of work.

Any latency and/or phase shift in the sensor will greatly limit its usability as a servo controller. From your traces, it seems to be lagging quite a bit, so it's either the speaker that's too slow, or the sensor, or both. Incresing the gain is supposed pull the system earlier, closer to the signal, but when you did, the sensor's trace started looking weird.