TOPIC: LED Modules and an Experiment in Red

I'm posting this to the Electronics DIY section because I think it would be helpful for people really looking to mod their sabers in-depth and because it outlines something I plan to work on where I may need some feedback. If you have any questions, comments, or concerns leave a reply, and if there are any issues on the location or content of the post please let me know so I may revise or remove it as needed. The discussion focuses on the difficulty of designing red LEDs but the principles can be applied to all colors.

The total will be divided into a few responses/sections for the sake of legibility:

• Introduction
• Module Components
• Module Construction
• Red LED Experiment

INTRODUCTION:

The Saberforge 12W LED module, as well as many others on the market, use Cree LEDs. These typically meet the current, voltage, power and cost demands used by most modern sabers. In most colors these function beautifully; however, when you get to the red spectrums, you’ll find that the wavelength of the emitted light falls between 620-630nm; about 10nm longer than the red-orange variety, 35nm longer than an amber color, and 30nm short of a ‘pure red’ 650-660nm. For further frame of reference, the green LED covers 520-535nm, blue falls between 465-485nm, and royal blue falls between 450-465nm.

Additionally, you must take into account the “relative spectral power distribution” of the light. The red LEDs here are shown to contain more light at frequencies below 635nm and less light at frequencies above. This means that the LEDs will look mostly red, but will also produce not insignificant amounts of red-orange light.

For these reasons we can see that the red LEDs contain at least some portion of orange/red-orange light. At this point it falls on the human eye and photographical technique for how the light is perceived- cameras, for example, may balance the light such that the blade has a much more orange sheen to it, or your eyes or brain may be more perceptive to the orange tones. Either way, some people will be completely satisfied with the red color (I personally like a bit of warmth to signify the fury of the dark side) or others may find the difference between the ideal white-core-red-glow to be disappointing. And the difference seems to become more apparent as the power ramps up.

Either way, one would have to ask why these LEDs are used in the first place if they do not meet the light enthusiast ‘standard’ for red color. The simple answer is there is a limitation to modern semiconductors that results in a gross inefficiency at the defined wavelengths. As such, it is generally accepted that with today’s technology a red LED will be red-orange. Look at most car brake lights or stop lights and you will see either red-orange clear-when-off LEDs or white LEDs filtered through a red plastic (which still appears partly orange when on).

650nm red LEDs do exist. The best alternative here would be the 650nm ‘Color Red’ Cree XPE LED- it runs at the required wavelength with mostly similar specs to the XB-D leds, is part of the line of LEDs currently being incorporated into Saberforge’s 12W+ module, and is produced by the same manufacturer. So wouldn’t it be easy to make a Photo Red module?

Electrically, the XP-E Photo Red LED is very similar. From a radiant/luminous flux perspective it is a bit… lacking. I seem to have misplaced my calculations since the last thread, but I remember it being an expected 40% brightness drop from the current red 12W module. This is my proposed shortcoming of the module- of course, I would prefer physical proof than just speculation and explanation to make a solid conclusion.

This has all been discussed in the previous threads, but I wouldn’t make a new thread without having some type of additional contribution to the community. I would like to treat this as a mind dump and a log for an experiment I want to conduct using the Cree XP-E Photo Red LEDs in a lightsaber.

THE PARTS
I decided to examine the possible outcome of making a 'pure' red module, and have been doing so in my free time for the past few weeks. The first step was determining exactly how an LED module is constructed.

THE LED:

LEDs can be thought of (functionally) as diodes. They only allow current to pass in one direction and have a somewhat constant voltage drop across themselves.

The voltage drop, in this case, is the key. To complete the circuit, enough potential to overcome the P-N junction of the diode must be expended. You can think of the voltage loss here as a toll- the circuit must pay X volts or no current can cross the bridge. There are equations you can use to calculate the exact voltage required from various parameters, but it is simpler just to grab the value from the tables in the spec sheet. Of course, the voltage changing will cause the overall current to change and therefore change the voltage which changes the current… but all we need is an approximation here. About 3-4 iterations should be enough.

THE POWER SOURCE:

Also important is the power source, and this design makes some assumptions:

1) The saber will be run from a 3.7V Li-Ion battery. This means, at its peak charge, the saber must be able to handle 4.2 V and still have reasonable performance at 3.7V.

2) The module is designed to run at 12W consistently, but can run over or under. In this case, the saber will be performing above 12W at full charge and below at low charge. The voltage should fall to a manageable level before the heat of overloading the module causes too many issues.

THE DRIVER/RESISTOR:

The next important parameter is the driver. In most cases, a simple resistor suffices. This is necessary because the LED only soaks up a finite amount of potential.

To illustrate: Your LED has a constant, invariable voltage drop of 2.5V (unrealistic, I know). Your power source generates 3.7V until it runs out of power. If you place the LED alone between the two leads of the battery you will be left with 1.2V of potential across the negative lead; this potential is not sunk into anything, so it instead must drop across the resistance of the wire (assumed to be incredibly small).

By Ohm’s law (V=IR) you just expended 1.2V over a near zero resistance, causing an infinite current (read: short circuit) through the battery, LED, and wires. This would either destroy your LED, destroy your wire, destroy your battery, or cause the battery’s PCB (it has one, right?) to shut it down. Not a good situation.

By imposing a finite resistance, you are allowing a path for the remaining potential to be spent and therefore allow your circuit to run successfully.

We now know the three things necessary to run an LED: A power source of sufficient potential and capacity, an LED to create the light, and a resistor to sink the remaining power. But how are they organized and interconnected?

The Spec sheet for the diodes tells us that the voltage drop is ~2.5V. We also know that the maximum voltage generated by our battery is 4.2V at full charge. This means, by Kirchoff’s Voltage Law, the potential is enough to drive just one LED in series. It must be concluded that each LED is in parallel with the others, with no two in series.

The big question is where the resistor(s) is/are located relative to the LEDs. Either each LED has one, the entire module has one in series, or they are otherwise connected such that each LED has a resistor ‘downstream.’ What’s the best way to tell the preferred design? To buy one, of course!

MODULE CONSTRUCTION:

Once my 12W Red LED arrived in the mail I decided to do a test of the brightness and color. To do this I attached the leads of the LED directly to the terminals of a NCR18650B battery. Immediately, a few things became apparent.

A) The module gets VERY hot VERY fast. I don’t know what I was expecting, but I thought it would take at least a little more time before the heat picked up. Luckily, in this case, all XB-D and XP-E LEDs have the same junction temperature, so the heatsink should not play a module role in the LED design.

B ) While the light does look a lot more red in person than in photographs, the core still does look a bit orange-y when held close to surfaces (I used a few different colored/textured surfaces to compare to).

C) While I was expecting the module to be either resistored with a single through-hole resistor on the negative lead or to be have 4 surface mount resistors in line with each LED, I found that there were actually two separately resistored leads that spliced together to form the negative contact. The resistors used on this component were 1 ohm, or an equivalent 0.5 ohms total (in parallel).

The design itself is very clever and interesting: by avoiding surface mount resistors the soldering on the board and risk of burning out the LEDs was minimized, and many other surface-mount design issues were avoided. Additionally, it means the resistors will not share the same heat sink as the LEDs and will instead radiate using the air or blade itself. The use of two resistors rather than one allows resistors of smaller power capacities and larger resistances to be used, reducing the overall size and cost of the components.

THE EXPERIMENT

Now that I had gotten a good look at the module, I decided to begin an experiment. What would happen if I tried to alter the module to use four Photo Red XP-E LEDs instead of the current XB-D units?

The process began with an analysis of the current and voltage across the device at a given time. From my findings, at full charge, the module will actually run at an average of 13W that will fall to about 9¼ W over the runtime of the blade. Of this power, roughly 65% is actually applied to the LEDs while the rest is applied to the two resistors.

To achieve similar numbers using the Photo Red LEDs, I determined the best resistance to drive the module would be 0.58 ohms with a total power capacity of >5.5W. Unfortunately, after browsing through Digikey and Mouser, I found only one brand that could supply the necessary resistors, and these had either a minimum order of 1000 resistors or a wait time of 10 weeks. Since I would probably lose interest by then, I needed to determine a more readily available resistance. After some trial and error and checking my numbers I decided on a total 0.6 ohm resistance provided through either two 1.2 ohm 3W resistors or two 1.2 ohm 5W resistors. While the 3W tolerance should be enough, I purchased sets of both just in case.

In all, the total power of the module should be reduced by about 0.4W at full blast and the power across the LEDs makes up about 6% less of that power than with the default design. However, it should theoretically drive the LEDs at a similar energy efficiency.

The big issue with the overall proposed module is the radiant/luminous flux of the two LEDs (higher numbers mean more light in this case). Comparing the luminous flux values at 350mA:

Red Cree XB-D: 67.2lm
Red Cree XP-E: 73.9lm
Photo Red Cree XP-E: 350mW; 0.350W*0.107*683lm/W=25.6lm

The Photo Red LEDs should produce a module around 60% dimmer than the standard 12W module.
To check my results, I have ordered the necessary components to modify a 12W module to see what this brightness difference will appear like in person. My first hilt that will accommodate the module will not be arriving until ASP goes through either, so a full in-saber comparison is still in the wings as well.

For reference I've attached a simplified diagram showing some of the expected values of the existing 12W module, on the left, as compared to the proposed design, on the right.

Superb thread. Some of the math is a bit over my head but I understand the basic idea behind the experiment. I wouldn't be the least bit surprised to see a 60% drop in brightness. Real true red light like the kind used in old film darkrooms is just not bright.

I recently got a champion Fallen in red and Im more than happy with it. Is there some orange in it? Sure there is. But the overall flavor of the LED is definitely red and its all the brighter for being shifted away from true red.

Well done Snakeeyz99, I look forward to all the results.

This should get a "sticky"

Well done..

Fantastic resource! Thank you so much for sharing this with us!

I attempted to solder the new parts to modify the LED module on New Year's Eve, and realized a bit too late into the process that I had forgotten to take LED size into account. After replacing the resistors and detaching the standard XB-D's, I popped open the XP-E's and realized they were nearly four times the size and would not work at all with the star on the standard module (I forgot to order 12W+ in my order details). Not only that, but I damaged the XB-D LEDs when I removed them so I can't even revert back to a standard 12W.

All that's left is to salvage what I have and try again This means I will either need to find a new star that can hold the XP-E's and try to recycle what's left of the 12W module or I can order a new 12W+ module and attempt the alteration that way.

This is not the first time I've made a simple mistake like this, and while I'm able to eat the cost fairly reasonably it's still annoying

Fantastic post!

Sorry to hear about your annoying mistake I'm glad you can salvage it.

I look forward to hearing your results.

Update: So I had my suspicion after seeing a 5 month old post just this morning, only to confirm it when my 12W+ modules arrived... 12W+ was an upgrade to Cree XQ-E LEDs, not Cree XP-E LEDs. XQ-E's are not available in photo red and are actually the same size as the XB-D LEDs.

This means an LED module design with photo red LEDs is much more design intensive than I would want. I would need to design or find a LED star with 4 times the surface area of the current model, I would need to obtain or design a lens that would focus the new spacing correctly, and I would need to pay to fabricate any necessary designs. Essentially, my idea is pretty much dead in the water. I'll still try to rig up something that will properly light up the XP-E diodes to compare them, but it will take a bit more time, research, and money that I had planned for.

I don't know where I got the idea that SF used XP-E for the 12W+ modules. I think the thread I saw it in was deleted.

Excellent write up thanks! Do you per chance have photos of all the dies you were using? I'm curious to see the variations between each.

Great Post!

I just took my 12W saber apart and measured the LED unit. To my surprise I measured only 0.8A at 3.2V for the entire package. This would result in the LED unit running at 2.56W instead of 12W. What's goint on here - any idea?

Oh, there were some replies...

I'll try to get some photos up tonight, but odds are this particular experiment is dead in the water. I also didn't realize you could buy a 3-Cree in photo red, which may work albeit not at a full 12W. Kouri has a saber that runs a photo red LED, so you may be able to dig up an example to compare to a 12W red.

Burney: What method did you use to measure the current across the module? Also, what tier and color of saber are you testing? Any detail you can provide would help.

I wouldn't worry too much about a Tri not operating at 12W - a Quad LEDs biggest advantage over a Tri isn't in brightness, but increased options for color mixing. There was a big discussion on some of the other boards when Quad Rebels became available, and the end result was they weren't much brighter than the available Tri-Rebels, pretty much due to limitations in the optics. It doesn't matter how much light your LEDs produce if that light's not being properly focused down the blade.

Also, instead of one set of resistors shared between the parallel LEDs, I think you might be better off wiring each LED individually to its own 1.2ohm resistor. 2W minimum would be safe, and TCSS currently carries 1.2ohm 3W resistors. That'll run the Photo Reds at 1A when the battery's at 3.7v, but Crees are usually good up to about 1.5A, so you'll be fine at the 4.2v of a freshly charged battery.

If you're willing to experiment with custom LEDs, there are lots of other things to try. Cree XT-E Royal Blues fit under XP optics and have a safe current of 1.5A, proving brighter than an XP-E2 Royal Blue. XP-L Whites are also blindingly bright with a safe upper limit of 3A, while also being available in single and Tri stars (have fun trying to keep 'em cool though). On my experimental to-do list is an XT-E Royal Blue Tri LED running at 4.5-6A inside a Photon Green blade.

You can also tinker with some custom Luxeon Rebels. The Cree XP-E2s are brighter in almost every color (Rebel Royal Blue is brighter on paper and a bit more purple in-person), but Rebels have some differences that make them interesting. They have single Cyan LEDs which are just what they suggest. Rebel Blues also tend to be a bit closer to Arctic Blue than the true Blue of a Cree.

Finally, Rebel Limes are amazingly bright (almost twice the lumens of a comparable Green). They appear yellow-green, but a more accurate approximation would be Not-Blue light, since they cover most of the Red-Green spectrum. They mix well with Royal Blue LEDs to make white light, and I've personally been meaning to try a rB/rB/L Tri-LED for a deep blue blade with a brilliant white FoC.