I came up with this idea, tested it and implemented it a while back, but I never got around to writing the article. So here’s the first new circuit article I’ve posted in well over a year. The basic idea is very similar to what Philips implemented with the Xitanium driver for Luxeon Star LEDs. It’s a current-regulated driver, and there are a few design advantages over a voltage-regulated resistive approach. And it’s cheap.
In response to reader feedback, let me also point out that this design is not optimized for efficiency, so it is probably not a circuit you would want to use to drive a Luxeon Star in a flashlight or other power-constrained environment. But for a lot of applications, this appropach works fine.
The basic theory here is to use an LM317T adjustable voltage regulator with a current programming resistor to get a constant current source to drive the Luxeon Star/O safely.
a simple circuit to drive Luxeons safely
The circuit itself is pretty simple, and is just a minor variation on an application found in the LM317 datasheet.
Here’s the analysis: It’s convenient to think of the LM317 as a 1.25V voltage source where pin 2 (Vout) is at 1.25V with respect to pin 1 (ADJ). The 3.9 ohm current programming resistor across that 1.25V source will always carry a current of I = 1.25V / 3.9ohms = 320 mA. If you connect a load between ADJ and ground, Kirchoff’s current law requires that the current delivered to that load be 320 mA.
I picked 320 mA as the target design current because it’s close to, but comfortably beneath, the rated 350 mA current for the Luxeon Star. If you want to adapt this circuit as a constant current supply for some other load, just do the calculation for R = 1.25V / the target current.
Power dissipated in the current programming resistor is a consideration at these load currents: P = 1.25V * 0.32 A = 400mW, so a 1/2W resistor is required for the 320 mA design. Power dissipated in the regulator will be dependent on the value of Vin, so be aware that a heatsink will likely be necessary to cool the regulator depending on your input and load voltages.
Input voltage is a consideration. The voltage drop across the load has to be less than Vin minus about three volts. The three volt drop is made up of the internal voltage drop across the regulator circuitry, plus the drop across the current programming resistor.
What this means in practice is that this design is perfectly suited to drive several Luxeon Star LEDs in series as long as you can provide a DC input voltage high enough to overcome the forward voltage drop of the series combination plus the three volt drop for the driver.
An example of that would be driving four white Star/O LEDs, the cumulative forward voltage is 4* 3.65V = 14.6V. Add 3V for the driver, and the minimum supply voltage is 17.6V. Also note that the LM317T is capable of accepting a supply voltage up to 35V, so this driver circuit could run up to eight Luxeon Stars in series.
If you shop around, the parts to build this should cost less than a dollar, but you can definitely build it for less than two. This design is superior to the use of a current-limiting resistor with a regulated voltage supply, since less power is dissipated in the resistor for most cases, and more importantly this can be used with a relatively poorly-regulated input voltage without harming the expensive Luxeon Star.
looking into the business end of a blue Star/O
and a red one
Alert reader cpemma provided this link to an outstanding tech brief at OnSemi which shows some variations on this theme for those of you wanting to build series-parallel arrays of Luxeons (or other high-current LEDs). The circuit on page two adds a few transistors and resistors but protects your array from overcurrent in the case where one of your parallel strings fails. First class circuit, and well-spotted, cpemma.
