Ok 12ax7
"I'm a tad confused, about the pick up for the feed back"... well...
On the right hand side the triangle is the speaker. R13 is the load resistor in series with the speaker coil, and so any activity in the transducer is reflected in the voltage across this resistor. This means the original drive signal, but also any backemf generated by the transducer.
By comparing the backemf to the input signal we can deduce an error signal, that can be fed back into the system and correct any anomolies that were being created by the speaker cone and it's resonant surroundings.
A copy of the input (what should happen signal) is fetched via R17 and fed to the + input of the U1B. A copy of the "what actually happened signal" from R13 is fed to the inverting input of U1B. These are compared and the resultant error signal is fed to the input amp U1A via the rc network through R8 into the inverting input, which corrects the input signal to allow for the errors found.
This all happens in real time, and so it can compensate for speaker location, resonance of the room, and poor speaker box design. I deliberatly built two boxes out of 1/8" plate steel with no damping inside, horn reflex... out of steel, just to give a terrible acoustic box... with this circuit, the bass was clean and clear, and the box sounded like a million bucks. Mid and treble were very lively, and no muddiness.
It is very useful learning device, particularly when used with the freq splitter, and three of these per channel..High,Mid and Low.
Hope this clears up some confusion.
Phillips had a MFB system back in the seventies, and they incorporated an accelorometer (microphone) and a fet amp built onto the cone itself to generate the "what actually happened" signal. It works well,but this system delivers the same result, without having to build a microphone and amplifier on every speaker you wish to use.
Some Sub woofers are apparently using some kind of MFB, but I don't know how they are doing it, microphone or backemf.
..............oztules