Thought I would bump this around again to hopefully learn more about damping factor and how it relates to different amplifier designs for MI applications.
Ok, here goes some information on damping factor. There are some specific things that I am going to be intentionally general about because it might get in the way of some IP and patents I am working on, but otherwise it should provide the information that you are looking for.
Damping factor is simply the load impedance (of the speaker) divided by the source impedance (of the amp, which is the amp's output impedance). The load impedance of a speaker is taken to be a nominal or weighted average but in actuality it varies quite widely with frequency (for an 8 ohm speaker, it can typically vary from 40 ohms down to 5 ohms). The output impedance of an amplifier (which is measured at the output terminals of the power amp) is typically quite low compared with the load impedance, how low determines how high the damping factor is. The output impedance is not a single number but a set of numbers that varies with frequency due to several aspects of how an amplifier's circuitry typically behaves.
One thing that marketing folks seem to grasp is either a super giant or super small number is better to manipulate a potential customer with, ESPECIALLY in the audiophile world. With distortion, vanishingly small numbers are proudly trotted across the adds and with damping factors the super high numbers are highlighted as though they mean something miraculous is going on. In fact, when you understand how these numbers are achieved it becomes obvious (or should) that in most cases the negative artifacts can be worse than the very numbers might suggest.
For the DF specification, the load is defined as a resistive load (either 2, 4, 8 or 16 ohms) and just the amplifier's output properties are explored. For a numerical example, an output impedance of 0.01 ohms (typical for a solid state amp) would result in a DF of 200 at 2 ohms, 400 at 4 ohms, 800 at 8 ohms and 1600 at 16 ohms. Now, due to a variety of reasons the DF falls with increasing frequency as negative back decreases (critical for stability), as the inductance of any filters (either Zobel or output reconstruction depending on amp class) begins to become significant, with output device gain falloff, dominant pole compensation, Miller compensation, etc. It is typical for the DF at 20kHz to be from 2 to 10x lower due to naturally rising output impedance.
Ok, why do these large DF numbers make no practical sense in the real world? Several reasons, the first being that when you add the source impedance of the connections and connecting wire between the speaker and the amp, in the best of circumstances this might add another .03 ohms so the system DF will end up being (8/.01+.03) = 200 rather than 800. Hmmm, this is an interesting bit of information that the marketing guys kind of omitted. Now, the other much more interesting issue is the real world speaker aspects, the easiest to understand is the amount of voice coil that sits outside of the magnetic gap's primary influence. Take a driver with a .5" gap height and a voice coil with a 1" winding height, this leaves (simplistically) 1/2" of wire outside the gap that does not have anything to do with motor force. If the DCR of the 8 ohm nominal VC is 5 ohms, than there's 2.5 ohms worth of resistive source in series with the part of the voice coil that is actually in then gap. If you were to do the DF calculation with this as part of the source impedance, you now have (8/.01+.03+2.5) = 3.2. (in reality, about 1/2 of that wire outside the gap is still influenced by the magnetic field because it does not stop abruptly) This also explains why efficiency falls with high Xmax drivers, all that wire outside the gap dissipates heat based on its DC resistance.
Another aspect of the DF effect is that when driving a real world speaker system, two factors come into play... first, the speaker system has mechanical damping (based on electro-mechanical filter models that I have discussed before) which combines with electrical damping to create total system damping. The other effect is that you also have large impedance peaks where the LF driver's impedance might reach 40 ohms, so the system DF will vary by a factor of 5 around these impedance peaks. For those of you who might be interested in modeling technology, this is what makes real world loads so hard to model accurately and realistically. There are many things happening all at once, just in the DF aspect. Now add distortion, frequency response, dynamic non-linearities and such and it's a very complicated system. I come from a mostly analog, industrial control system background where all of these things must be considered when developing a working negative feedback control system with servos, actuators and proportional controls, so the basics of this stuff kind of becomes second nature.
In actuality, very high damping factors really are insignificant BUT the circuit attributes used to create these high numbers can easily contribute to large negative artifacts such the the cure is far worse than the presumed disease.
I hope this helps with the understanding of WHY sometimes numbers aren't quite what they seem.
