Nuts and Volts - January 2008

BY FERNANDO GARCIA

■ FIGURE 1. Light output vs. junction temperature.

The first challenge was to remove the excessive localized heating which causes a loss of luminous efficiency (shown in Figure 1) and may even set off thermal runaway. Additionally, LEDs are diode junctions, and require their current to be regulated. This is simple, trivial resistor stuff when only a few milliamps are required. Not so trivial — and quite wasteful — when dealing with high currents.

A possible solution would be a current regulated switchmode power supply. However, switchmode supplies do not tolerate excessive current swings, especially when going from zero to full blast and again back to zero in a millisecond. (Hmmm ... who said this would be a simple project?)

To top it all off, these LEDs have a Lambertian light distribution; “Lambertian” being a fancy word meaning “all over the place.” Some sort of lens or reflector would be required to form a useful beam.

First things first ... which LED source should I employ? After reading countless datasheets, I settled for Luxeon’s K2 in a “star” configuration. As shown in Figure 2, the LED die is attached to an aluminum star-shaped plate, which itself is to be mounted to the heatsink. This plate also provides the necessary solder pads to attach the wires.

I quickly determined that an all-aluminum box would both house the project and serve as a heatsink. I could then use hardware (similar to what is used to mount TO- 3 transistors) to attach it to the aluminum box. But this hardware caused an interference with the optical lenses. As discussed above, the LEDs require lenses to focus the light, and as shown in Figure 3, these are attached to the LED assembly with an adhesive pad.

A possible solution would be to employ an adhesive thermal co pound, but I wanted something that is widely available to hobbyists. A workable solution was found: Apply a tiny bead of silicon thermal compound in the center of the star (on the flat side that will attach to the aluminum box) and push the star firmly against the aluminum box’s face.

If done correctly, the bead will spread out evenly and NOT smear out. This is important, and it make take a couple of trials to get it right.

Then, cyanoacrylate superglue is employed on the star’s periphery to perform the actual bonding. Done correctly, this provides a good thermal interface and fairly secure attachment.

optical, and mechanical aspects of the assembly, then came the fun part. How to drive these LEDs? There are literally dozens of high output, constant current switchmode LED drivers available from different IC vendors. However, they either are in nightmarishly-tiny SMT packages, or optimized for steady state output and/ or a single light shot such as a flash.