Let’s Build An LED Lantern! Part 2: Assembling The Light Source

Hi all. In part one, I laid out my idea for the lantern’s basic design; what I want it to do, and what I plan to build it out of. Today, I’m going to begin putting the thing together, starting with its light source.

CYA stuff: While no dangerous voltages will be directly present in this project, it will be using large lithium-ion battery packs, and those can be very dangerous if mishandled. It also involves soldering, which can expose you, obviously, to high temperature molten metal, but also to potentially toxic vapors from burning flux. If you do decide to make one of these yourself, you’re doing so at your own risk.

I’m going to start with a little theory on LEDs, what they are, how they work, and why I’m using them. If you didn’t know, LED is short for ‘Light-Emitting Diode.’ A diode is the simplest semiconductor device you can build: It’s a single P-N junction that acts like a one-way check valve; current can flow through them in one direction only. An ideal diode has infinite resistance to current flowing in the wrong direction, meaning zero reverse current flows through it, and in the case of an LED, no light gets emitted. (In reality, all diodes will either start to conduct in the wrong direction or get fried if you pump enough juice through them.)

Technically, all diodes are light-emitting, since they all convert some quantity of the electricity that flows through them into heat, which is radiated as long-wave infrared light. But that’s just good old-fashioned black-body radiation. The silicon dies of LEDs contain special dopants which emit nearly perfectly monochrome light when electrons pass through them, in a process called electroluminescence. These dopants determine, in most cases, the color of light the LED emits, from infrared, through visible light, and well into ultraviolet. Electroluminescence is the reason LEDs are such energy-efficient light sources compared to incandescent bulbs, and why they’re ideal for this project: Unlike incandescent lamps, LEDs don’t waste energy burping out light at frequencies we don’t need. A typical LED used for lighting will emit roughly 200-300 lumens per watt of energy consumed. By way of comparison, an incandescent lightbulb with a tungsten filament emits about 16 lumens per watt on average.

All diodes have what’s called the forward voltage, abbreviated VF. For regular diodes, the forward voltage is the minimum voltage that has to be present across the diode before it will “turn on” and start conducting electricity. For LEDs, VF usually represents their typical operating voltage, and like I mentioned earlier, it’s a function of the color the LED emits.

Another, more important quantity we need to take into account is the maximum current. LEDs are current-driven devices, meaning that small changes in the current flowing through them have big impacts on their brightness and their power consumption. In fact, an LED will happily consume all the current you can feed it, right up to the point where it overheats and dies, even if you keep VF in spec. To keep that from happening, LED manufacturers document a maximum and/or nominal forward current rating for their LEDs, abbreviated IF. Unlike VF, which is largely constant among LED colors, IF varies wildly with LED types and applications. A dinky little indicator LED, for example, might have an IF of 20 milliamps, while an LED used for lighting might have an IF of an amp or more.

Basically, all this talk about voltage and amperage boils down to the fact that the power supply I use must not supply too high a voltage to my LEDs and–more importantly–not source more current than they can handle. To get an idea of what those parameters should be, let’s take a look at the specs for my LEDs. In part one, I mentioned that I’ll be using three of these Cree XM-L RGB+W LED modules in my build. Here’s the datasheet from Cree’s website:

According to the sheet, the red chip operates at a nominal VF of 2.25 V, the green operates at 3.3 V, and both the blue and white operate at 3.1 V. As an aside, the blue and white LEDs have the same VF because they both use blue LED chips. “White” LEDs almost always contain one or more blue LED chips covered with a yellow phosphor, which absorbs & re-emits some of the blue light. The primary emission from the blue chip mixes with the secondary emission from the yellow phosphor to create white light. The thickness and composition of the phosphor layer determine the white light’s color temperature: The thicker the layer and/or the more orange the phosphor, the warmer the white light is tinted.

Anywho, these LEDs can operate at up to 1 amp per die or 4 amps per package, according to the datasheet. That’s not atypical for high-brightness LEDs, but that is still a ton of current to push through a semiconductor, and more importantly, try to dissipate as heat. (2.25v + 3.3v + (3.1v * 2)) * 4 amps = 21.7 watts! LEDs will of course die immediately if they get too hot, but their service life will also be dramatically shortened if they’re run close to their max ratings for too long. To try to keep that from happening, I’m going to underrun these suckers, and shoot for an IF that’s closer to the 350 mA nominal rating per die.

OK, enough theory. Let’s get building! I’ve soldered five wires to an LED unit: One to the negative (cathode) solder pads for each color, and a fifth wire that connects all the positive (anode) pads together. I’ll get into the significance of this a little later when I discuss controlling the light, but for now, what I have here is an LED chip that will light the appropriate color when the positive side of the circuit is applied to the black lead, and the negative side is attached to the matching color wire:

Here it is in action:

Incidentally, the LED is connected to this Arduino Uno board, a preview of things to come:

The four colored cathode leads from the LED are connected to four output pins, while the common anode lead is connected, thought a 1,000 ohm current-limiting resistor, to a 3.3 volt power source on the board. The microcontroller is running a very simple program that pulls the first output pin LOW to ground potential, thus completing an electric circuit and lighting the first LED color. It waits one second, sets the output pin back to HIGH to turn the LED off, sets the next pin in the series LOW to light the next color, and so on. Again I’ll go into all of this in more detail in the next article.

Now I need to gut the donor bulb I mentioned in part one and use its housing as a support & heatsink:

With a little effort, the light diffuser at the top snaps off:

Removing the three philips screws and desoldering the wires in the center caused the both the plate containing the LED chips and the ballast circuitry inside the bulb to fall right out. In a bit of cosmic serendipity, the screw holes lined up perfectly with the indents on the new LED chips, allowing me to screw them right down without drilling out more holes:

The grey goop underneath the chips is a thermally-conductive grease. The grease fills microscopic gaps between the heatsink and the LED modules, allowing them to thermally couple together much more effectively. This is a critical step, unfortunately omitted by the makers of the original bulb.

The next step is to wire the three LED packages together in series. In a series circuit, the power source and the LEDs are all connected together in a big loop. The source feeds the first LED, which feeds the second, which feeds the third, which sinks current back to the source:

There are four of these circuits pictured here; one for each color. The finished product is a bit ugly, due to the limited space I had to work with and the fact that the heatsink was doing its job sucking heat away from the soldering iron, which made it tough to get the solder to melt.

The last step here is to snap the diffuser back into place and test the unit. Since the LEDs are wired in series, I need to feed each circuit three times the forward voltage it takes to light an individual color. Red needs 6.8 volts, blue and white need 9.3 volts, and green needs 9.9 volts. That’s easy enough to pull off with a bench power supply, so let’s plug them in!

I have to say I was pretty impressed with the results. The light from each LED chip diffused fairly evenly out of the bulb, and the look of an intense monochromatic light in a darkened room was very striking. Seeing it officially lit up for the first time gave me encouragement that this project would not only work out but the end result would be worth the effort.

That’s it for part two. In the next article, I’ll dive into controlling and regulating the LEDs. Thanks for reading!