The Arcmania SureFire KT TurboHead towers are available in kit form from AW (search for "tower" in the Dealer's Corner in the CPF Marketplace). $44 each including shipping IIRC.
You then have to add your own emitter, 0.55" (14mm) driver board, thermal epoxy, wire, solder, and manpower.
I've built multiple AW towers using an SOB 1000 driver and a Seoul P4 U2-bin LED. At 1000ma drive current, a P4 U2-bin is rated at about 260-285 emitter lumens. IIRC, I get about 13,000 lux at 1 meter, vs about 18,000 lux with an N62 lamp in a SureFire 12ZM T-62 TurboHead. The thing is that the 18,000 lux only lasts for a short time. Then the incandescent lux starts heading steadily downward as things heat up and the batteries get drained rapidly (within minutes). The stock 12ZM is really impractical. In contrast, the LED tower just chugs along in constant current regulation, generating very little heat. I can run the tower using 2x123A primaries, 3x123A, 4x123A, 2xLi-ion, or 3xLi-ion -- basically from about 5V to 16V for full output -- which gives me a lot of flexibility. I can also run with 1xLi-ion and still get a lot of light out of the tower.
I also built an interesting Badboy Nexgen/Seoul P4 tower to run at low voltage (e.g., 2xNiMH). I would estimate about 200 emitter lumens with this setup, which is clearly brighter than if you tried to run an SOB 1000 at 2.4V.
I just completed a tower using an SOB 1000 driving a 2S2P K-bin, WC-tint Cree MC-E. At 500ma drive current per die, the K-bin MC-E is rated at around 500 emitter lumens. I haven't done any extended run tests yet to evaluate heat generation, nor made any measurements of hot spot lux or tailcap current draw. TBD. So far, I've run the MC-E tower using 4x123A, 2xAW16340, and 2xAW17670. Beam pattern in a T-62 TurboHead shows a typical MC-E "cross" at close range (say 6' and under). It is unclear if I have the optimum tower height for the MC-E. With an AW "Seoul focus" tower, you have to raise the MC-E emitter by an estimated 0.94mm. I used an 0.03" ( about 0.76mm) thick copper shim and hoped that the two thermal epoxy layers (tower/shim and shim/MC-E) would make up the rest. I need to get my calipers out to measure the added height. At longer distances, there is some slight "texture" to the hot spot. It is noticeable only in white wall testing. The rest of the beam pattern is excellent. No dark rings. A very bright, smooth spill. The MC-E hot spot is wider but not as bright as the P4 hot spot. The MC-E spill is much brighter. Overall MC-E output is much higher. When I use ceiling bounce illumination in a room roughly 20'x15'x10', it's like I turned on a ceiling light.
I use these towers in an old SureFire 12ZM. Before building these towers, my 12ZM was a shelf queen. Now I actually use it. I also have a P4-based tower in an M900 VFG weaponlight with a KT4 TH.
I have an SOB 1200 on order and plan to see how that works in a tower.
At these power levels for driving an MC-E, I believe you should pot the driver board in the tower. You don't need any fancy thermal management if you use an SOB 1000 to run a Seoul P4.
