What I have not seen or discussed anywhere on the web is the mixing of these styles in series for crossovers. Eg, solid core with an air core trailing it. Is there any benefit or detriment to this scheme?
Yes, that could be done, but under very specific circumstances with may not, in practice, ever occur.
Observation: series inductance adds, much as how parallel capacitance adds. So it is possible to fabricate larger values from smaller ones.
Consider,
argumentum, a contrived situation where:
(a) a combination of acquisition cost, size, mass, and DCR constraints preclude a fully air-core solution
— but —
(b) sonic requirements for distortion arising hysteresis and/or saturation preclude a fully iron-core solution.
A hybrid would therefore be required to satisfy the constraints of this contrived example better than either type alone could.
It would therefore be
possible to move to the middle of cost/weight and distortion (DCR, hysteresis, and saturation), by obtaining some of the inductance from an air-core inductor (higher cost, larger size, greater mass, higher DCR) and some from a iron-core inductor (lower cost, smaller size, lower mass, lower DCR).
This would, of course, be predicated upon the bedrock axioms that the trivial DCR of the air-core inductor actually mattered for sonic quality and that the distortion from the iron-core correspondingly did not have a significant effect upon sonic quality. Such assumptions would be mistaken.
This is a gedanken experiment and no one would in practice perform it. Hobbyists can scrape together the few dollars to move to air-core, and companies are generally so focused on price that iron-core would be a requirement to reduce overall cost, with the only choice being how much distortion to add, as the core type is a cost knob.
Also, if a third toroidal inductor is physically placed, perhaps vertically, between the other two inductors, does that put up a wall to protect all from electromagnetic fields?
Magnetic materials couple, never shield each other.
Special shielding materials are required to block magnetic fields. Look up µMetal or mu-Metal which was widely used in high-bandwidth CRT monitors and televisions to shield the gun from stray magnetic fields. Mu-metal is difficult to work with, as it must be annealed after welding or soldering.
RTally is totally correct about separating inductors, as the field strength varies as 1/r^2.
Also, if one discovers too much filtering by an inductor, could a resistor be connected prior to and after the inductor (semi-bypass) to change it's crossover point?
RTally is, again, totally correct about the purpose of a filter: it removes signal and separates the frequency band such that each driver receives the appropriate signal.
Your posited situation of wanting to attenuate a low-pass filter is generally improbable because the inductor is a
low-pass filter, so it is
intended to
remove certain frequencies at 6 dB per octave. Such as with a woofer or as the ceiling for a band-pass filter used with, say, a midrange. So attenuating the filter would permit unwanted signal in the stop-band to enter the driver. This would not be desirable.
Permit me, however, to rephrase the question to one which is not nonsensical to pose.
Suppose,
argumentum, that
impedance compensation is desired to slightly flatten the driver's response in a
very specific frequency range. A network of components, using either L or LC, is therefore required to shape the driver's response so as to make it better conform to the flat ideal.
Such a compensation network is, of course, placed in
parallel with the driver and itself functions as a filter, albeit one which is limited in intensity for if it were not it would remove the entirety of signal in the pass-band. So the compensation network must be constructed as either a low-pass, high-pass, or band-pass, except that unlike the corresponding crossover filter for that frequency band the compensation network adds a resistor or voltage divider to
substantially limit the current
bypassed through the compensation network (in this case,
argumentum, LC or L), because to fail to include the resistor or divider would act like a band-stop with the usual rolloff properties. So certain frequencies would be attenutated, but not removed in the traditional means. The resistor or divider would therefore limit the current through the inductor.
Attenuated filters are not generally used outside of compensation networks, as the only value would be to flatten response in a narrow region.
