For any of you who might be familiar with the CentraLED, check out these renderings of our prototype flashlight:
http://bpond.smugmug.com/gallery/2024704/1/103777449
This flashlight uses the same lamphead as the CentraLED, which has a flood beam pattern providing smooth, even illumination of an area. Note the knurled ring at the base of the lamphead. Twist that ring for continuously-variable output.
The picture is obviously just a CAD rendering, but we had working prototypes on public display at our major automotive trade show (SEMA) in November '06. We then showed production-ready models at SHOT '07. A few CPF'rs came by our booth at SHOT and they seemed suitably impressed.
I would say that the mechanical engineering required to pull this off far outweighed the electrical engineering. The design uses a potentiometer, whose axis of rotation needs to be common with the flashlight's central axis. Once a pot is chosen, that defines where the rest of the PCB is located and creates all sorts of design constraints. Now you've got to get battery power in to the PCB, and LED power out. One set of those connections is spinning with the knurled ring and the other set is stationary with the flashlight body. It's not trivial to do this.
Electrically, I'll be the first to admit that yes, it uses PWM. We put a lot of effort into finding an optimum PWM frequency to minimize the strobing effect, and I think the results are pretty darn good. The pot is a component of a very simple timer circuit which fires the enable pin on the driver IC. It's not rocket science; pretty much every driver IC datasheet teaches you how to do this. Electrical complexity with this design is quite low and therefore reliability is high. We looked first at doing current control instead of PWM, but the problem was the lack of availability of an appropriate potentiometer. Continuously variable current control requires the pot to provide the feedback signal directly to the IC, and to achieve maximum brightness at one end of the scale, the pot resistance value had to be very nearly zero. Pot's just aren't built that way.
We are using a twisty tailcap with a tactical momentary, and we are establishing a small fixed current on the low output end so that no matter where the ring happens to be positioned, you'll get
something when the light is turned on. It works out to about 6 lumens on the low end and 80 lumens on the high end. Power source is 2xCR123A. Initial production will all be black HA-III, but we are looking at offering a camo model later. Output adjustment is easily done one-handed, with the body of the flashlight gripped in the palm, and thumb and forefinger twisting the ring.