Here is a schematic for the NANJG 105C. Let me know of any errors or omissions and I’ll update the pic or add corrections to this comment. Notes: When looking at the battery side of the driver, the stars are number from right to left. +Vin is the spring in the middle, -Vin is the ground ring on the edge of the driver.
There is a lot of effort going on to improve the technology of LED (Light Emitting Diode) lighting. The reason is that LEDs are much more efficient than standard incandescent or CFL light bulbs. They also last longer than either of the alternatives and don’t have issues with disposal like CFL bulbs. LED lighting has a lot going for it except one thing. It is still pretty expensive. At any rate I have been doing a fair amount of reading in the engineering trade journals on the subject and have been wanting to play around with LED lighting. Over the last year or two I have been collecting high output LEDs that I have been able to pick up inexpensively at hamfests and surplus places. In some case I have gotten some nice LEDs for under $.05 each.
When you are using an LED as an indicator in a project you usually just drive it with a voltage or microcontroller and a current limiting resistor. In these situations you really don’t worry too much about driving the LED for its maximum output or worry too much if the current changes a bit due to variations in the supply voltage or changes to the VI (voltage-current) curve as the LED heats up a bit. You also don’t tend to drive it anywhere near the LED's limits so there is plenty of margin and the LED will last practically forever.
In a lighting application you want to get as much light as possible out of them so the LEDs should be driven with constant current. The same amount of current should always be going through the LEDs regardless of changes in the circuit over time, temperature or other reason. Needless to say, there is a lot of work going on developing integrated circuits for driving LEDs. Most are similar to switching power supplies. These are very efficient, often over 90%. The downside is they are somewhat complex for playing around with and many come in fine pitch surface mount packages that make building a circuit without a circuit board a major exercise in frustration. Many are kind of expensive in small quantities as well. My pile of inexpensive LEDs continued to grow. Finally I came across a neat circuit in the January 6, 2011 edition of EDN magazine. The author, Eliot Johnston, used a simple two transistor, two resistor analog constant current circuit to drive LEDs in an outdoor illumination system. Brilliantly simple! Sorry about the pun.