2nd-order filter phase confusion

Stickjam said,

"Please help me understand what this article is saying... "

I think I understand stickjam's confusion. There are two considerations essentially, the polarity of the drivers as they are wired, and the phase shift that is inevitable in the crossover region. With both drivers in the same polarity, a two-pole butterworth crossover will be 180 deg. out of phase at crossover creating a null. If they are opposite polarity there will be be a peak in the response at the crossover frequency, yet they will sum. The Q 0.5 of LR alignments eliminates the peak. If it is a two pole LR, then the polarity must be reversed between the two drivers in order to sum at crossover. This is not minimum phase. With a 4 pole LR, the drivers may be of equal polarity and sum at crossover as well. This is ideal.

7th order filter - could this form the basis of a decent phase shifter?
Stephen

[quote author="StephenGiles"]7th order filter - could this form the basis of a decent phase shifter?
Stephen[/quote]

For a decent analog time delay you typically want a response without overshoot, hence something like Bessel. A lot of stages are needed to get a decent wideband delay.

For producing a fixed phase difference between two outputs, stagger-tuned all pass networks work pretty well. The typical design achieves a 90 degree difference over the audio band, tailored for single-sideband modulation apps.