Subwoofer X/O frequency experiments

The important factor is a subwoofer that doesn't conk out at 100 Hz...

True, especially to avoid sub drivers' distortion as the frequency goes up.
Generally speaking, I've noticed that most of us use subs as woofer complements, i.e. to offer extra energy that's lacking in the 90-40Hz region -- rather than to introduce energy in the 16-40 region.
The difficulty with most subs is, drivers cannot reproduce both (say) 16Hz & 300Hz (about 4octaves!) signals at the same time w/out some form of distortion and major difficulties in blending with the sats. They hit mechanical & electrical limitations. With the sub set high the resulting sound in the room may "sound" pleasing at first, but, the reproduction anomalies quickly show up in classical music (to cite one example).

Gregm...What you say suggests that the subwoofer should really be two-way, 20 to 50 and 50 to whatever, and I agree with this. Each of my three subwoofers consists of two drivers, 15" and 12" driven by separate power amps, and I have actually tried the two-way subwoofer thing. It doesn't seem worth the trouble, so I use both drivers through the full SW range so as to take advantage of maximum cone area to minimize cone excursion. Perhaps because the cone excursion is much less than most SW, my drivers (remember there are six of them)don't have the IM distortion problem which you describe...not to any noticable extent.
"Blending" with the main speakers is also not a problem even with the higher X/O frequency. My SW are located directly (2 ft) behind the MG1.6, and off the floor so that their sound goes through and around the Maggies, just as if the Maggies were producing it.

For the most accuracy you need your sub-woofer configured such that its sound sums flat with the main speakers at your seating location (this requires the appropriate magnitude and phase relationship)

For the most accuracy from any speaker you need it placed so that you're not getting cancellations from the reflected waves as the difference between direct and reflected sound approaches a 180 degree phase shift (282 feet to the point of reflection/ frequency) and so that the bass/midrange increase you get as it couples to the wall behind it is addressed with its cross-over baffle step compensation.

For the most accuracy from a system you need to avoid pushing drivers high enough that they break up or beam, or low enough that the distortion that goes with excursion is not too high.

The type of music, playback level, etc. may change which distortions are less objectionable but can't change physics.

Transfer functions are all that matters whether they result from acoustic or electrical effects.

An analog low-pass or high-pass filter or digital filter that acts the same (Signal processing is outside my area of engineering) has an inherent phase shift. The outputs from high and low pass filters of the same type and order put the outputs 90 degrees out of phase for each order.

The highest bass accuracy also results when you don't put energy into a room below (565 feet / longest dimension) Hz.

All this explains why big speakers don't work well in small rooms and what sub-woofer/main configurations produce the flattest (or most enjoyable with a given type of music) response.

If you have a sealed speaker its mechanics mean you have a 2nd order roll off. You get it and the woofer to sum flat with a 2nd order electrical low-pass on the sub that inverts its signal, the F3 points match between speaker and cross-over, and both share the same Q. This is why sealed speakers mate well with something like an REL that has a cross-over with adjustable frequency and Q.

The attenuation and phase shift an octave into a system's pass-band are negligible. You can high-pass a set of speakers an octave above their F3 point and low-pass your sub at the same frequency. They'll sum flat with a fourth order electrical cross-over. Or a second order electrical with the sub inverted. This is why fourth order cross-overs an octave above the speaker's F3 point work well.

In a sealed speaker excursion doubles for each octave lower (halving of frequency) you go. Once a ported speaker starts rolling off its excursion quadruples with each octave drop. Distortion comes from excursion, so distortion increases faster in ported speakers below their cross-over point. A 4th order cross-over maintains constant excursion on the speaker so excursion isn't increasing. This is why a higher order cross-over works better with ported speakers.

When you pair a sealed speaker with a second order high-pass and sub with fourth order low-pass you can get them to sum flat. THX home theater systems do this with sealed speakers having poles at 80H Q=.707, a second order butterworth electrical high-pass, and fourth order Linkwitz Riley low-pass. You have an LR4 response on both.

When you get a 1/2 wavelength length difference (say 3.5' from a wall for 80Hz) you get cancelations. With a speaker 4' out you don't have cancelation above 80Hz. With the sub-woofer in a corner you don't get a cancelation at any frequency. By the time frequency drops enough for a speaker to couple to the wall behind it it has little output. The sub-woofer is always operating in the same space so it is unaffected by this. This is why softit or in-wall speakers can work very well.

A higher order sub-woofer low pass keeps its cone out of breakup.

Etc.