I'm posting this first try because it was like a well designed exercise in what not to do...
The satellites are simple 2-way systems consisting of a small woofer and a tweeter (typically referred to a WT, short for woofer and tweeter) These systems were built using:
A pair of 0.25 cubic foot enclosures from Parts Express for about $120.
A pair of Tang Band W5-704D mid-woofers, about $25 each from Parts Express.
A pair of ND20FB-4 tweeters, about $8.00 each from Parts Express.
Passive crossovers, designed using in-box measurements with SounEasy.
The speaker terminals and crossover parts were about $30 each from Parts Express.
All drivers were flush-mounted.
The W5-704s are good drivers. They can be crossed over quite high (compared to many 5" mid-woofers). The relatively soft poly cones don't have any bad break-up problems to deal with and some have used them up to 4 kHz. I listened to them full range for several days (the tweeters installed, but not connected) and I must say, they weren't half bad.
The little ND20FB-4 tweeter begins to show increasing non-linear distortion (harmonic distortion, producing sound that isn't part of the original signal) as they go below 3500 Hz. They are usable a little bit lower, but will sound cleaner (clearer) if used above 3.5 kHz - 4 kHz.
Both drivers are tested and reviewed on Zaph's site (tweeters) (5.5" woofers) , and both are quite good for what they cost.
My first attempt was pretty bad (horrible actually). It wasn't a fault of the drivers, enclosures, or electronics, it was my fault. I got carried away with trying to get a smooth system response and wound up trying to cross to the tweeters at 6500 Hz. I had mounted the tweeters as close to the edge of the woofers as possible, but the center to center spacing is still 3 7/8". The frequency of a sound wave corresponding to this wavelength is 3495 Hz.
As the crossover (XO) frequency is increased above 3.5 kHz, the wavefront generated by both drivers operating together at the crossover frequency develops areas of reinforcement and cancellation. This is commonly referred to as "lobing" or in bad cases, a comb effect. The center (or primary) lobe of the sound field generated by both drivers begins to resemble a torus or doughnut, becoming smaller (thinner top to bottom) with increasing frequency. Above and below the primary lobe are the first two areas of cancellation, then two more lobes caused by reinforcement, etc, etc.
When these lobes are viewed in cross section, looking at the edge of the baffle, you can see what I'm trying to describe. For a fixed ctr to ctr spacing, the higher the XO frequency, the thinner the primary lobe becomes. In extreme cases, there are many lobes above and below the primary lobe.
Moving the XO frequency to 6500 Hz was an extreme case.
As I moved my head up and down I could hear the sound changing. The relatively fat primary lobe resulting from a crossover frequency of 3500 Hz or lower, was reduced to a very thin lobe with many repeating bands of reinforcement and cancellation above and below it. The best description I've heard is that the vertical polar response looked "like a bunch of bananas".
As I moved my head vertically, I could hear each lobe. This didn't require much movement either, an inch up or down was enough to make the sound change. For any fixed ctr to ctr spacing, the thing to remember is that the vertical polar response of a pair of drivers operating at the XO frequency is frequency dependent; The higher the XO frequency, the more lobes.
The crossover is what controls this problem between two drivers in a speaker system. As the signal moves away from the XO frequency, one of the drivers' output (relative to the other) will be decreasing. With a typical LR4 (Linkwitz-Riley, 4th order) acoustic response from the drivers, either drivers' response will be down 24 dB one octave away from Fc (the XO point). For an Fc of 3.5 kHz, the woofer is 24 dB down by 7 kHz, the tweeter's response is down 24 dB by 1750 Hz, and the interference between the drivers is nil.
Since the phenomenon increases at higher frequencies, it is particularly important when placing the tweeter relative to the woofer or midrange driver operating below it. It is seldom a problem when placing a woofer relative to a midrange driver in a 3-way system (when the speed of sound = 344 meter/second, at 1000 Hz, the wavelength = 13.54", at 500 Hz, the wavelength = 27.09").
For a system using flush-mounted drivers, crossing over well below the frequency who's wavelength corresponds to the ctr-ctr spacing will yield a fat (tall) primary lobe, and the sound won't change with vertical head movements. On the other hand, the thing that usually limits how low you can cross over to the tweeter, is the rising harmonic distortion tweeters exhibit with decreasing frequency.
Remember, as you push a tweeter below a certain frequency (look for frequency where its HD climbs above 1.0 %):
The worse it will sound.
How much distortion can you listen to?
You may be surprised, 1% - 2% may sound fine, in some cases, you may not notice it. On the other hand, you may be able to hear HD levels as low as 0.1%. Try the tests on the Klippel site.
The less power it will be able to handle.
As you lower Fc, the tweeter has to work harder, this will involve higher voltage signals and more current being passed through its voice coil. It's relatively easy to damage a tweeter's voice coil.
How loud do you like your music?
How expensive will it be to re-cone/replace your tweeters?
This is in an important portion of the sound spectrum, where most people are still very sensitive to distortion. There are still a few fundamentals occurring at 4 kHz, and many sibilant sounds that harmonic distortion will tend to magnify. This becomes much more important/noticeable when the crossover point occurs in the 1.0 kHz to 2.5 kHz range.
After a couple of weeks, I finally realized what I had done and I redesigned the XOs for an Fc of about 3500 Hz. This and the construction details are described in my second try.