thinking of building 7 way loudspeaker

[steph_tsf]

Originally Posted by MisterTwister

What I want to do is speaker with one driver per octave (roughly) and no direct radiators. It will be very efficient, relatively small and inexpensive (no need for high end drivers, since each driver operates in narrow passband)

Would be interesting to see this working. Especially if miniaturized. I like your idea consisting of using 4th order bandpass enclosures for the 125 Hz and 250 Hz units, with some acoustic gain provided by the enclosure. Do you know the drivers you are going to use for the 125 Hz and 250 Hz units ? Are you planning to use a passive crossover approach ? Expect suprprises when using passive crossovers. You need to simulate them, along with the enclosures. LTspiceIV can help you doing this. Coils for crossovers at 125 Hz and 125 Hz are heavy and expensive : can you cope with this ? Or will you only high-pass filter the 125 Hz and 250 Hz sections using a capacitor in series ? Keep an eye on the impedance : if you only high-pass the 125 Hz and 250 Hz sections (for avoiding coils), they will eat current even outside their bandwith. And thus, the overall efficiency will be poor. You may try a series filtering arrangement instead of the usual parallel filtering scheme. Or, you may use an active filtering scheme, but that's not cheap : you need then N separate power amps, plus low-level active filters to be carefully designed (and simulated). For the subs, what kind of enclosure are you going to use ? Ready-made subs including an amplifier ? For the subs, if you don't use ready-made subs including an amplifier, you may prefear a custom-made 8th order bandpass for maximal efficiency. Keeping the same philosophy at all stages. I see that you may want to use high efficiency horn loaded speakers for the 500 Hz, 1K, 2K, 4K, 8K units. You need to determine if the directivity pattern of such drivers are compatible with your expectations. And, even if they are, you need to determine if the crossovers (phasing) are going to ruin the directivity pattern (using simulation). I don't think you'll end up with something satisfactory, but I guarantee that if you pay attention to all the points I raised, you'll quickly become an expert in loudspeaker design.

[twest820]

Originally Posted by panomaniac

Can you explain a bit more? My experience with the DCX has shown that any channel can be anything you want. Have you found otherwise?

I don't have a DCX, but according to the manual it implements what Linkwitz refers to as the wrong topology for proper management of phase. My experience is in good agreement with Linkwitz's remark that phase interactions between channels aren't too much of an issue with three ways but become problematic with four way and up.