The main thing is, we don't want it to be higher than other mechanisms of distortion. Once the contribution of current distortion is below that of, say, suspension non-linearity, the focus needs to shift.
It's a nice video and makes me appreciate the engineering sense of Dali.
I thought this worth sharing from Mr.GPT:
Question: Is speaker input current distortion proportional to linearity of L(x)? (Inductance vs. excursion)
Answer: In principle, yes — to a first approximation, distortion in the input current of a speaker is directly related to the nonlinearity of the voice coil inductance L(x). But the relationship isn’t purely proportional; it depends on how L(x) interacts with the signal and other parameters.
Here’s the breakdown:
The voice coil impedance is
Z(x, i, ω) = R_e + jωL(x, i).
When L(x) varies with cone position, the instantaneous inductance changes with signal amplitude (since x follows the input current).
This variation causes current modulation because the back EMF depends on d(L(x)i)/dt. As the cone moves, the inductance changes, altering the effective impedance.
The result is amplitude-dependent current, i.e., current distortion.
In small-signal terms, if you expand L(x) as
L(x) = L0 + L1x + L2x² + ...,
then distortion terms in current are approximately proportional to these coefficients (L1, L2, etc.). So yes, more nonlinearity in L(x) means more current distortion.
However, it’s not strictly directly proportional, because:
Other nonlinearities (Bl(x), Cms(x)) also affect current indirectly via back EMF.
The effect depends on frequency — at low frequencies, L(x) distortion is negligible compared to Bl(x) or Cms(x), but at high frequencies it dominates.
The shape of L(x) (symmetric vs asymmetric) determines which harmonic orders appear.
So, in short:
Speaker input current distortion increases with the nonlinearity of L(x), but not in a perfectly proportional way — it’s one contributor among several, and its influence grows with frequency and signal level.
