There are plenty of projects that use either the mid-woofer or the tweeter, but I'm not aware of one that uses both in this configuration.

The result is a very smooth and balanced speaker, easy to listen to, and I don't feel I've lost any detail or accuracy over the RS180/RS28A 2-way's that these replaced: a win-win from my perspective. This is a viable combination of drivers.

The crossover is from jAy (jkim). Jay is intense and passionate about precision; he has produced a topnotch crossover for this design. Many thanks to Jay for working on this, for all the modeling he did and advice he provided during the many times I was headed in the wrong or unproductive direction.

My own crossover modeling was based on incorrect impedance data, both for the mid-woofers and the tweeter. I have since gone back and determined where I went astray. The lesson learned was to get impedance plots of sufficient resolution and/or visually match the impedance curves generated by T/S parameters with measured impedance (e.g. by playing with the Le Coeff. and Le Expon. values in Jeff B.'s Frequency Response Modeler tool).


Objectives
A primary motivation for this project was to replace a pair of metal driver-based speakers (RS180/RS28A MT 2-ways) that were highly detailed and accurate, but otherwise tended to be harsh and somewhat fatiguing.

I also wanted to branch out into areas or aspects of speakers that represented new territory for me, i.e. MTM configuration vs. MT, towers vs. standmounts, building and finishing the cabinet vs. purchasing pre-made/custom-built.

Drivers
The criteria for driver selection were:
- non-metal cones/domes; either paper, poly or composite materials
- measured and evaluated by other DIYers; I am without a measurement set up.
- low to mid-range in cost
- easy to work with and would not require a complex crossover
- good cosmetics - the baffle would remain exposed with no grill cover

Crossover
Fc is at 2-2.1KHz. Many have employed a conjugate filter to tame the XT25's impedance spike. Jay and Augerpro argue that the filter is not needed in all cases, particularly with higher order networks and appropriate crossover point. In any case, the 30 Ohm resistor in the high pass helps mitigate the effects of the spike.

After voicing for several weeks (and subsequent orders from parts merchants), I landed at the low end of the Jay's range for the primary cap, at 10.1 uF and slightly higher on the secondary cap, at 25.2 uF, on the high pass circuit. The low pass is per the diagram.


Cabinet
Baffle width is 8.5", with .5" roundovers. Driver spacing is 5.5" center-to-center.

Tower volume is 1.55 cu. ft., not accounting for driver displacement or bracing. The port is 3" in diameter, which gives it a very wide appearance, and flared at both ends, so no port noise of any kind. End to end length is 5.75", including the flares.

The top, back and sides of the cabinet are lined with either eggshell foam or sonic barrier. The bottom of the tower has polyfill stuffing below the port.

Deployment/Impressions
I have these in a 2 channel music system with both digital and analog sources; both sound great and natural. They are driven nicely by a NAD 325BEE, and are quite efficient as you would expect from an MTM. My room is 12' by 36' and there is plenty of SPL. The front firing port allows for placement closer to the back wall than what I would do for rear-ported speakers. This helps with bass response, which is deep and tight, but not loud or forceful. No sub needed though, in my opinion.

UPDATE: attached is a crossover diagram with the final component values I settled on after voicing.