Recently, I had the experience of re-manufacturing the tweeters in a pair of speakers I had bought. Due to a "senior moment", I fried the tweeters in a pair of Quadral Chromium Style 50 speakers, and the local dealer decided to f*ck me over even though I admitted the mistake and paid for replacements. Why they decided to screw with me is another story, no doubt involving elements of the RCMP who would prefer not to be named. At any rate, I was left with a pair of drivers with unequal efficiency, and it annoyed the living snot out of me.
So I decided to take the two broken tweeters and rehabilitate them. If you look into the Quadral speaker line, you will discover that the tweeters use a filament suspended between two powerful magnets in series with a 5.1 ohm resister. The filament is apparently a piece of aluminum coated with a thin layer of titanium, or something to that effect. My replacement is somewhat thinner, and required the manufacture of unique tooling to shape.
The result has less mass and consequently improved the sensitivity of the driver, with obvious implications to the resultant sound quality.
My amplifier is an Onkyo M282, which is apparently an amplifier that uses negative feedback, and made in Japan. On the Web, there are a number of articles which articulate the controversy over NFB amp designs, with numerous manufacturers shunning the technique for various reasons. None of the descriptions or reviews take into account the most important factor in an audio system -- the interaction of the speakers with the amplifier.
I believe I have achieved such good results primarily because I reduced the "mass" of the tweeter, which in combination with the amplifier design, has helped produce a superior sound. The key to understanding the thinking behind this is the interaction of the amplifier with the speakers.
Speakers are an inductive load, AFAIK, which means they present some of the same problems as electric motors in the design of driver circuitry. In the instance of an audio signal, the drivers have a hysteresis contingent upon their ability to accelerate in the presence of a electrical impetus. In my nascent understanding of electronic circuits, this resistance to movement will interfere with the amplifier output in the same way a naive power supply will drop-out under load.
The negative feedback circuitry in the amplifier will detect this drop-out and drive its input harder as a result, thereby reducing perceived distortion.
In the available literature there is talk of the "delay" that results from the signal excursion that occurs from amplifier distortion before it is corrected. I would suggest that this "delay" is imaginary for the reason that the negative feedback circuitry in fact operates essentially instantaneously, and applies a continuous correction.
It is worth noting that distortion at the amplifier output is a result of two sources: firstly there is the distortion introduced by the gain stages in the amplifier, and secondly, distortion introduced by the mass of the drivers, which interact with the magnets and impose an inductive signal on the amplifier output. In either case, the negative feedback circuitry should nearly instantaneously apply a corrective factor to the input signal which will effectively coerce the drivers to conform to the shape of the input signal -- if the NFB circuitry is not too weak or slow.
My suspicion is that by reducing the mass of the tweeters in my system, I have relieved a burden from the NFB circuitry in the Onkyo M282, and thereby improved the resultant sound quality by an order of magnitude or so.
The apparent controversy over the effectiveness of NFB amp designs seems to ignore the amplifier-speaker relationship, and the role it plays in fidelity. An ideal speaker would have no electric mass, and would therefore have zero influence (or load, I guess) on the amplifier, and would obviously reproduce the amplifier signal perfectly. Loudspeakers must necessarily have mass, if only represented by the air they displace, and will therefore affect the amplifier by virtue of inductance. Consequently it is my feeling that negative feedback amplifier designs represent a good solution to the practical problem of signal reproduction.
In thinking about this subject, it has occurred to me that a NFB amp will perform better if the speakers it drives are more expensive than another, all things being equal. Even ambient noise in the listening environment might affect the negative feedback circuitry.
Perhaps obviously, I have no test equipment to use to measure the effects or phenomenon that is at issue. Just my common sense. But if you were here, you would be able to attest to the fidelity of the ultimate result.
Comments and constructive criticism is welcome.