This is the one that takes 8 D cells, and has two Cree XR-Es. http://www.amazon.com/Coleman-200000.../dp/B001TSCINE This is mostly an electrical review, as the other aspects have been covered elsewhere.

I have one of these, and after doing some measurements, felt compelled to take it apart and see how it works Here's what I found:

The batteries are configured as two banks of 4, for a nominal 6 volts. Each bank passes through a 1N5819 diode before connecting in parallel; I'm not sure if its just to drop the voltage, or for reverse protection for the LED. In any case, this is the first weakness; the Vf of the two diodes are inevitably mismatched, and so the lantern pulls harder on one bank than the other. I measured 0.94 amps on one bank and 1.057 amps on the other at full brightness, with the imbalance increasing at lower settings.

The switch selects a position on a resistor ladder, so there are zero to six additional resistors placed in series. The XR-Es are each in series with a 1.1 ohm resistor and in turn wired in parallel.

This is a completely resistive design, so it is pretty inefficient; just estimating from the voltage drops (6 to 3.7), you'd get about 39% of the battery power lost in the resistors. Also, on NiMH batteries, you can't get full brightness because of the lower voltage.

The cases is reasonably moddable. The resistor ladder and diodes are on a small board behind the switch, and easily accessible. The top can also be removed but with difficulty - It twists off but seems to be tacked with glue so I had to pry it off the first time. There is plenty of room on the top and it is vented.

I started trying to do a better regulator, based on this: http://www.edn.com/article/488339-Sw...ent_source.php This turned out to be a poor design for a flashlight or lantern, because the op-amp really needs a +/- voltage supply. I got it to work on the bench, but it needed + and - 12 volts or so, not really suitable. Also, that regulator was taxed at 2 amps (two XR-Es), and I'd like some more headroom since it will run at full power most of the time. I did at least learn some about how it worked. The op-amp shouldn't need a two-sided supply, since the voltage drop across the current sensing resistor will always be positive, but I don't know how to take advantage of this.

Here's the request: Where should I look for some better designs? I'm sure this has been done many times, but I haven't found many suitable circuits.

Some considerations: I'd like to keep the dimming feature (seven steps of dimming). I also have the option of wiring the batteries and LEDs in series instead of parallel which should fix some of the balance problems in the original design, and also lower the current through the regulator.

Thanks!