Spectrographic plot
Spectrometer plot of the white Luxeon Star I LED.
Ocean Optics USB2000 Spectrometer on loan from TWO-CUBED.

Spectrographic plot
Spectrometer plot of the white Luxeon Star III LED.
Ocean Optics USB2000 Spectrometer on loan from TWO-CUBED.

Spectrographic plot
Spectrometer plot of the white Luxeon Star V LED.
Ocean Optics USB2000 Spectrometer on loan from TWO-CUBED.

Luxeon Star Warm White /O LED, (
(Rec'd 12-15-2003, tested 01-03-2004)

This is the long-awaited 1.2 watt "warm white" Luxeon Star /O LED, now being sold by such outfits as out of Australia. (They ship all over the planet, so don't let their location throw you off).

The /O means the LED comes with an acrylic optic, mounted to the Luxeon Star PCB using a cylindrical black plastic spacer. This convention follows other Luxeon products, so if you see a /O after its name, it comes with the optic and mounting cylinder already in place. The optic appears to be a NX01 type, so the LED inside is probably a low-dome (batwing) type.
(Edit next morning around 10:36am)
I popped the optic off, and sure enough, the LED inside is a low-dome type. The phosphor appears to be a light, slightly brownish yellow, rather than the light whitish yellow you might see in cool white Luxeon Star LEDs.

This Luxeon Star is mounted to a square MCPCB (metal core printed circuit board), rather than the hexagonal board that other Luxeon Stars may be mounted to. This board measures 1.0" on each side, and has two notches in opposite corners so that it may be affixed to a heat sink.

close up of Luxeon Star
Here is a picture of a white LS /O so you know what the LED looks like.
Photo courtesy of Lumileds, and was used with permission.

The color temperature is significantly warmer (yellower) than most other white LEDs, probably coming in at around 3,000K. Overall, this LED has a warm, yellowish white color to it.
The brownish color at the perimeter was caused by the camera, and is not at all present in the LED beam. The camera does not always do well for beamshots, and here is one example.

Measures 164,200mcd with a test current of 364mA.
This measurement is consistent with what I know about the warm white 1.2 watt Luxeons.
The bin number on this sample is MP4JW.

Luxeon Star 5-Watt (Lumileds); scheduled for 3Q 2002 release; some models may only be released to OEMs)
Last updated: 08-24-02
First, there was the 1-watt Luxeon Star. It was BRIGHT. Now Lumileds is in the process of introducing a 5-watt version of this eye-stabbingly bright LED! I know very little about it, other than it appears to use 4 discrete emitters coupled together in a series-parallel arrangement inside the device; yet it takes up no more physical room than the original 1-watt Luxeon Star.

Closeup of 5W Luxeon emitter
This is a photomicrograph of the four emitters on their heat-dissipating slug.

Closeup of 5W Luxeon emitter
And this is a photomicrograph of the emitter assembly by itself, shown at even higher magnification.

Loan of the 5W cyan LS used in these pictures was made by "PercaDan" of Candlepower Forums.

A number of samples of these have reached developers, and I borrowed one of them to take these pictures. I did *not* however, attempt to operate the device anywhere near its rated capacity, as I was not able to solder wires to it that would have allowed me to carefully monitor its current. So I connected it to a current-limited power supply that delivers only 200mA at its maximum setting. I have just two hands, and both were being used to hold the power wires to its solder pads. I'd have needed another pair of hands to take a measurement under these conditions. :(

A prototype flashlight has already been constructed using one of these devices. I was able to borrow that as well, and will post pictures of it and the beam profile analysis I ran on it in the near future.

Luxeon Star 1.2 Watt (Lumileds); price/availability variable (see below); some models possibly OEM availability only)
Last updated: 10-20-03

close up of Luxeon Star
close up of Luxeon Star
Pictures courtesy of "evan9162" of Candlepower Forums, and were used w/ permission.

The above pictures show the truncated inverted pyramidal (TIP) die of a 1.2 watt red-orange or amber Luxeon Star LED. The output of one of these is blindingly bright, so you don't want to look directly at one with the naked eye while it is on.
The blue ring in the bottom photo is a reflection of a 1W Luxeon Star LED in a Mag-Lite flashlight that was used to supply the light for these shots.

close up of Luxeon Star
A white Luxeon Star /O (with integrated optics).
Photo courtesy of Lumileds, and was used with permission.

Spectrum of a white Luxeon Star with optic.

The Luxeon Star Power Light Source is the newest generation of very high flux LEDs. Manufactured by Lumileds (,, these breakthrough LEDs represent the best and the brightest LEDs you can get today.

First appearing as prototypes in the late 20th century, they are just now entering production and becoming available to end-users (that's people like you and me) and to manufacturers of new, energy-effecient, ultra modern lighting products. Full production is expected to kick in sometime in late summer 2001.

close up of Luxeon Star

This is a very close-up look at a white Luxeon Star, model LXHL-MWIA (also known as the Star/C).

close up of Luxeon Star

A selection of Lumileds Luxeon Star Power Light Source LEDs - some destined for museum history.
Some of these are prototypes from as early as late 1998 to as recent as late 1999. Very few people have these particular experimental models. Now you know why they're going into a museum. :-O

One of the samples I have is so old it still has the reverse-biased red LED chip inside the package, designed to indicate a reversed connection and protect the expensive InGaN chip at the same time. This has apparently been replaced with a zener diode in some InGaN units, and nothing at all in the more robust AlInGaP (amber and red emission) models that really don't need that type of protection.

There are very few units with the reverse-biased indicator/protection LED inside - if you have one, consider yourself very privileged & lucky. Don't mess with it or intentionally reverse-bias your LS at anywhere near full power as the red indicator will be destroyed immediately. It will only withstand a brief application of reverse-biased DC; just enough to protect the LS chip against a short-term "accident"; a one-time only reverse connection.

close up of Luxeon Star
A close-up of one of the /C units. Photo courtesy of Lumileds and used with permission.

Early red & amber models used a chip nicknamed Barracuda; more recent offerings use the Prometheus and Bluefish chips. These names are largely used only in the lab (like Atari's cute internal names it gave to its game consoles in the 1970s & 1980s), but among LED afficianados, that's what we call them in the bar too. :)

close up of Luxeon Star
Here is closeup of another LS model with the 'Bluefish' InGaN chip visible inside.
This one probably glows green, blue-green, or blue when energized.
Photo courtesy of Lumileds and used with permission.

The finished product goes under the commercial name Luxeon Star Power Light Source, or Luxeon Star for short.
As I understand, they are trying to discourage the use of "Prometheus" and other internal names.

Comparision of a LS and a LW400 flashlight

On the left is a Lightwave 4000, a flashlight using 10 overdriven Nichia white LEDs.
On the right is a single Luxeon Star, being driven at 220mA (it is supposed to be driven from 350mA, but my old power supply doesn't go that high). Note the substantially warmer color of the Luxeon.

This particular unit is rated at between 18.1 lumens and more than 23 lumens at 350mA!!!
Color temperature for this specific sample is listed as 5000 to 6000K; some productions can be as cool as 3500K or as hot as 7000K; Lumileds tests each one and seperates them into 7 different bins of related color temperatures - much the same way Nichia bins their blue LEDs for wavelength.

Measured values for all test units at 200mA and using the /O optics package hand-fitted over the LED.
Correction for ambient light should be: listed value of LED minus 1608.

Red: 75,000mcd
Amber: 70,000mcd
Blue-green: 46,800mcd
Blue: 33,000mcd
White /O: 63,100mcd
White /C: 64,800mcd

All of the above measurements were taken on a Techtronix J-16 with J-6511 sensor probe (optimized for/calibrated to the photopic curve); greatest accuracy will be with the Star /O in this batch.

Spatial distro curve
The average spatial distribution curve for the Luxeon Star /O in white.

beam profile analysis
A beam contour analysis with point illuminance readings for each zone.
The unit tested to 24.2 lumens with an input current of 395.400 milliamps.
Both of these charts were made using the ProMetric system, from Radiant Imaging.

These devices are NOW AVAILABLE to OEMs, and in some cases with a rather stiff minimum quantity requirement, to the general public as well.

Visit Future Electronics ( to order these.
Also, Arc Flashlight LLC has been given the green light to sell a limited quantity of these to individuals. Bulk purchasing is probably not allowed (this source is only for hobbyists who need only one or two!), go to http://www. to find out more. Regardless of where you purchase, expect to pay $14 to $26 each, depending on the seller, and the LED type (the /O version with optics costs more; the /C version on a bare MCPCB costs less).
As of early 2002, it is now possible to purchase the bare emitter (Level 1 assembly) without the MCPCB. These are also available through Future Electronics.

Contact info for Lumileds is as follows:
Lumileds Lighting, LLC
370 West Trimble Road
San Jose, CA 95131
Americas/Canada Toll Free Tel: 1-877-298-9455
Europe/Asia-Pacific Tel: 1-408-435-6044
Fax: 1-408-435-6855

A second white Luxeon Star (the /C version without optics) that was provided as an evaluation unit has a much higher color temperature than the first. When mounted above my computer monitor and used as a task light, the color appears as though I am working under a mercury vapor discharge lamp that has an arcing electrode in it. My keyboard is illuminated in a ghostly, violetish white light (more violet and less blue & green than an Hg lamp) as if there were a funky'old tyme' streetlight fixture outside my window. :-O

Color temperature is very high; estimated somewhere between 9,000 and 12,000K.
This is definitely one of the "bluest" "white" LEDs I've yet come across.

To use these LEDs, the following hookup instructions are recommended. (In this example, for the white /O unit).
1: Use a DC current source power supply, not voltage source.
2: Set current to 350mA, forward voltage will be 2.5-4.0 V.
3: Connect positive lead as indicated by copper dot on the MCPC board.

For sustained operation, please ensure that the unit is properly mounted to a secondary heatsink. The back surface of the MCPCB will stabilise at 60C without any additional heatsinking.

Strange quantum behavior in GaN and InGaN LEDs:
Gallium nitride LEDs can do strange things when you let the input voltage sag too low. Most of the time, it is never noticed, but it is very noticeable when it happens to a blue Luxeon Star LED.

Look what the BLUE Luxeon Star does when you feed it around 2.2 volts!
It isn't blue anymore.

As far as I can determine, what happens is that at low voltages, impurities within the lattice structure of the GaN chip have a narrow enough bandgap to emit low energy photons like yellow and red, but the wider bandgap responsible for the emission of blue light doesn't have enough voltage across it - so no higher energy electrons get across to recombine and no blue light is emitted.
Once the voltage across the LED is high enough for the electrons to "jump" the bandgap used for blue (for which the lattice structure is optimized) , higher energy (blue) photons are generated within the lattice structure in great quantity, which very quickly "overpowers" the dim yellow emission - so that at normal currents, all you see is blue. The original low energy yellow is still there, but it is masked by the blue which is emitted thousands or even millions of times more intensely.

Experimentation has shown that the weak yellow emission never reaches wavelengths shorter than about 578nm; so there is no real threshold between yellow and blue (green is skipped altogether). If the forward voltage is just so (and the range the red, orange and yellow emission occurs is very narrow) the yellow and blue emissions can be "balanced" in such a way the LED appears to glow white. Not brightly, but bright enough to see in moderate room light.
And although it's certainly nothing useful, it is an interesting piece of quantum behavior in GaN that very few know about. Now you too can be added to the ranks.

Here are some measurements I took:

"White": Vf=2.24 volts, If=566uA (this is the blue threshold)
Yellow: Vf=2.16 volts, If=200uA
Amber: Vf=2.11 volts, If=119uA
Orange: Vf=2.07 volts, If=55uA
Red: Vf=1.87 volts, If=24uA

'BARRACUDA', CIRCA 1997 (Lumileds)
Last updated: 10-21-01

NOTE: The following text article describes the LED that ended up being used in the Luxeon Star as detailed above, and the bulk of it was written in late 1999.

The Barracuda is a high-flux, high current red LED manufactured by Agilent (formerly Hewlett-Packard). NOTE: This is now a Lumileds product. Lumileds is a joint venture between Agilent Technologies and Philips Lighting.
Packaged in a very nonstandard looking case (see above!), this LED appears built for extremely high current and high temperature operation.

Barracuda LEDs
Two of the bare LEDs. These are supposed to be mounted on a
metal core PCB, and are almost never seen in their feral state.

Unfortunately, I know relatively little about the early 'Barracuda' design; other than what I can deduce by observation and tests.
Newer examples of this technology appear at the top of this page; this text is considerably older!

Barracuda LEDs
Notice the unusual star-shaped wirebond on the chip, and the unusual mounting.

The picture above shows one reason you can tell these are meant to be operated at high currents - the asterisk-shaped wirebond structure on the chip was designed to distribute a large amount of current evenly over the face of the LED's die or emitting chip, allowing the device to obtain brightness levels that are unprecedented and simply not allowable with "normal" T1 3/4 style LEDs. There is also a large amount of metal surface area designed to pull heat away from the chip. A metal plate on the bottom makes it possible to mount this to a heatsink (see Luxeon Star above) and drive it with several hundred milliamps.
The emitting chip itself is also many times larger than that of a "normal" 5mm red LED.

Both the Barracuda and Prometheus LED packages are filled with a liquid or gelatinous silicone compound, and not an inert gas as had been originally thought. The silicone formula is there to reduce thermal expansion stresses on the bonding wires. The dice are intended to have 280+mA run through them with proper cooling, and they do become quite hot.

The silicone is also closely matched to the refractive indices of both the LED die and the transparent LED case; possibly resulting in greater overall efficiency.
In the laboratory (and under some very certain circumstances only), this LED has broken the 100 lumens per watt barrier. Currently, the most efficient LED commonly available today just barely nudges 35 lumens per watt and only then occurs with a small percentage of any given lot.

When operated at "normal" LED currents of 20 and 30 milliamps, this LED is surprisingly dim - but when I cranked up the current very briefly to near 150mA, the LED really started to brighten up. Much more testing is in order here before I can post some brightness estimates, continuous current rating, and other parameters.
However, since these particular samples are not mounted on their special thermally conducting boards, I can never operate them anywhere near their maximum ratings. For that type of testing, I will use the Luxeon Star modules you see at the top of this page.

The successor to the Barracuda, called Prometheus (see note below), is apparently in current, but very limited production (summer 2001)..
It too is found only as a Level 2 assembly (already mounted to a metal-core PCB) and never appears by itself like most other LEDs do. Expect devices using this LED or something similar to appear en masse starting in mid 2001. An example of such a device appears here.

NOTE: As of summer 2001, Lumileds is trying to discourage the use of "internal" names like Prometheus and Barracuda.

They prefer the descriptive terminology be determined by the type of beam the LED package produces: the old Barracuda is now known as the "Batwing", and the Prometheus is now known as the "Lambertian". These terms describe the beam profile as it is seen on a beam profile analysis chart, and is more technically accurate than the names of saltwater fishes.

These parts can be distinguished from one another by both the physical appearance of the die and the lens profile. The Batwing has a cubical die and a short, almost flat dome lens; while the Lambertian (pronounced "lam BER' shun") uses a TIP (truncated inverted pyramidal) die and has a distinctly tall, dome shaped lens.
As of this writing (10-21-01), the Batwing configuration comes in all visible colors, and the Lambertian comes in only red and amber.

WHITE 5500-6500K InGaN+phosphor 
ULTRAVIOLET 370-390nm GaN 
BLUE 430nm GaN+SiC
BLUE 450 and 473nm InGaN
BLUE Silicon Carbide
TURQUOISE 495-505nm InGaN
GREEN 525nm InGaN 
YELLOW-GREEN 555-575mn GaAsP & related
YELLOW 585-595nm
AMBER 595-605nm
ORANGE 605-620nm
ORANGISH-RED 620-635nm
RED 640-700nm
INFRARED 700-1300nm
True RGB Full Color LED
Spider (Pirrahna) LEDs
True violet (400-418nm) LEDs
Agilent Barracuda & Prometheus LEDs
Oddball & Miscellaneous LEDs
Programmable RGB LED modules / fixtures
Where to buy these LEDs 
Links to other LED-related websites
The World's First Virtual LED Museum
The Punishment Zone - Where Flashlights Go to Die
Legal horse puckey, etc.
LEDSaurus (on-site LED Mini Mart)

This page is a frame from a website.
If you arrived on this page through an outside link,you can get the "full meal deal" by clicking here.