Here are a couple of shots of the Delta 15 main modules:
The woofer is a high quality prosound unit with remarkably smooth frequency response, which extends up to 4 kHz on-axis. It has enough linear excursion capability to not need a protective high-pass filter if you don't want to use one, up to about 115 dB or so. Beyond that a protective filter starts to make sense, though it's still arguably optional up to about 120 dB.
By buying and measuring the compression drivers a case at a time, we're able to generate matched pairs. Not that the quality control isn't good to begin with (these drivers match up more closely than most we've worked with), but we want good stable imaging, and matched drivers contribute to that.
You can see the rear-firing tweeter on the back of the cabinet, at the top. Here is the idea behind it: The waveguide-style horn on the front is pattern-matched to the woofer in the horizontal plane in the crossover region, but the woofer still has a wider pattern in the vertical plane. So there is a slight mis-match in the power response. We address that by adding some additional rear-firing high frequency energy, closely matching what the front horn is doing as far as tone, but reduced in level. The reduced level allows us to still place the speaker very close to the wall. You can effectively turn the tweeter off by slapping a piece of painter's tape over it, but you'll find that the sound now localizes at the speaker a bit more, instead of being detached from it.
Next you can see a resistor in a terminal cup. That resistor is used to tilt the front-firing compression driver's response up or down; in other words, it has a lot more effect up at 10 kHz than it does down at 2 kHz. At lower values (resulting in a more tipped-up response) a fair amount of power goes through the resistor, so I recommend paralleling two (12-watt Mills) resistors to get the value you want if you plan on playing really loud. So if room acoustics or amplifer characteristics or personal preference leads you to prefer a different top end than I would have picked, you can easily get there without using a scratchy, low-tolerance L-pad.
The ports I use are 3" diameter flared Precision Ports. Yes some midrange energy escapes out the ports, but because of the flares is doesn't sound harsh like midrange energy through straight-sided port tube does. Combined with the output of the rear-firing tweeter, the speaker is almost like a weak dipole, as you do get fullrange sound out the back. If you need place the speakers close to the wall or in a corner, you can close one or both ports using a plumber's expandable test plug (hardware store item).
The bevelled-off rear corners help give you visual index for the recommended 45 degree toe-in, and reduce the width of the footprint with such toe-in. The idea is to get the speaker axes to criss-cross in front of the listening position. With speakers having fairliy aggressive, uniform pattern control, like this one, such a configuration gives a very wide sweet spot. What happens for off-centerline listeners is, the far speaker is actually a bit louder because you're on-axis for it but well off-axis of the near speaker, and this compensates for the earlier arrival time of the near speaker. The secret is for the off-axis response of the near speaker to fall away smoothly and fairly rapidly. That's why I use big cone woofers and 90-degree waveguide-style horns. Another advantage of this setup configuration is that the first early lateral reflection of the left speaker's output is the long-path bounce off the right side wall, and vice versa. This contributes to a sense of envelopment. That being said, best imaging for one is with speakers aimed straight your ears. In this configuration you might find the top end a little hot with the normal tweeter setting, so you may want to back off a bit via the resistor cup.
What you don't see is the crossover. The speaker's target on-axis response isn't ruler flat, and this is because although the goal is "constant directivity" (and the horn we use is billed as such), the system does have some variation in the radiation pattern, and we take that into account. Also, we find a slightly downward-sloping in-room response curve to sound more natural than "flat". You can get back closer to "flat" by cranking in a bit of treble boost via the resistor cup on back. Our goal is approximately flat perception (erring slightly on the warmish side), not approximately flat by any one measurement.
Also not seen is an internal damping technique borrowed from architectural acoustics, and credit to studio designer Jeff Hedback for his consultation services.
