Current Drive for Loudspeakers

There seems to be a dearth of information on practical ways of current driving loudspeakers, though I think I have looked at all that is available from Hawksford to "The natural way of driving loudspeakers". What is clear is that you get 20dB or more of reduction in distortion, quite apart from the distraction of thermal compression.

I have an idea of putting a capacitor in series with the "current sense" resistor, so as to roll off the current drive when it approaches the fundamental resonance of the system, and thus have the nice predicable voltage drive through that region and not, there, be affected by the change in impedance. But otherwise have current drive above there, even though that would mean a rising response with impedance.

It seems to me that the capacitor would only have to be a few hundred uF, so not fall into the trap of real power capacitors, but yet be quite good in that the ESR will have fallen to about as low as it gets for any electrolytic. Although this isn't the topology I think I would like to use in the end, though I'm not quite sure what that is right now, it nevertheless encapsulates the idea in a simple form.

I am well aware of the dependency on impedance that this whole thing implies, but was wondering if anyone thought this was a bad idea in principle. The idea is simply a transition from current drive to voltage drive when we get to the point that matters. (And incidentally I don't think I believe the very generously smooth curves that Nelson Pass published in his study of various Fostex and Lowther units with a high impedance drive. I think a very much more real peak should be apparent in measurements; and one that reflects the transition from a Q of 0.7 to a Q of perhaps 3 or 5, which is what it would be without electromagnetic damping.)

Incidentally, but quite pertinently, I don't like chucking away perfectly good signal in a power resistor, so bright ideas of how not to do that, but do the same thing more efficiently would be amazingly welcomed.
Last edited: