by Chris Long and Mike Groth VK7MJ.

Almost since its inception, atmospheric optical communication has been limited by the impracticality of available hardware. Lasers are relatively expensive and are hazardous to eye safety in a populated environment - even if one could overcome the basic incompatibility of coherent light with a turbulent atmosphere. Transmit a coherent light beam through the atmosphere, and moving air cells of variable density will cause the beam's wave fronts to lose their phase coherence within a few hundred metres. This results in severe fading, scintillation and noise at the receiver, an effect analogous to the 'granular' appearance of a laser beam seen via specular reflection. Professional optical scientists and mathematicians, like Dr Olga Korotkova of the University of Central Florida, have attacked the problem by using thin diffusion filters over the laser source to reduce the degree of beam coherence. Her scientific paper on partially coherent beams mathematically analyses the problem:

Non-coherent light sources collimated through large fresnel lenses are less subject to these shortcomings. Until recently, hardware to generate and modulate a powerful non-coherent beam was very limited in speed, efficiency and intensity. Most gas discharge or arc lamps, for example, are unstable, non-linear, or provide their output at more than one discrete frequency with a limited intensity.

Around the year 2001, a new type of light emitting diode was commercially launched, representing a major breakthrough in power dissipation and light output. For the first time, LEDs became fully competitive with some incandescents for lighting. 'Luxeon' LEDs also have an incredibly fast rise time, the red (630 nm) Luxeon reaching full brightness in around 15 to 25 nsec. Their light output is fairly linearly related to input current with no hysteresis effects, and they can be monochromatic over a band 20 nm wide, to the 50%  flux level. They're stable and robust in operation. It takes a lot of electrical mismanagement to destroy them with current surges. Above all, when the Luxeon's output is spread into a broad beam by a fresnel, it's absolutely eye-safe. This non-coherent radiation isn't legally defined as a 'radio' link requiring a license. Anybody can use this type of link for transmitting any type of program material, regardless of mode, content, bandwidth - or supervision.

In a source area of less than 1mm2, the red 1 Watt Luxeon of 2002 could generate about 35 Lumens of output at a 2.95 volt, 385 mA input. Since then, the output of 1 watt red Luxeons has been improved to around 50 lumens for the same power input.

It immediately occurred to the authors that this was a near-perfect light source for atmospheric modulated light communication. In association with a large fresnel lens, a Luxeon with a Lambertian radiation pattern can throw a broad beam with a small dispersal angle, a situation ideally suited to transmission through the turbulent atmosphere. With this in mind, the authors imported some of the first red 1 watt Luxeons to reach Australia. By October 2002 we set a Tasmanian state optical comms DX record of 30 km, and with the same Luxeons in February 2005 we set the world audio-modulated non-coherent light DX record of 167.7 km. At the time of writing (June 2005) this record still stands. Refer to the following web pages for details:

Modulated Light DX

Six weeks after our world record DX contact, Lumileds released a new, enlarged version of of their red Luxeon, quadrupling the flux available from their 2002-vintage Luxeon 1. The new 'Luxeon III' (the red version was only released in April 2005) produces about 150 Lumens at 1.54 Amps and 3 Volts input. Admittedly, the source chip in the lambertian Luxeon III has increased to over 2.5mm2 so that its intensity is only (approximately) doubled, but its source size makes a better optical match with the focal inaccuracies of the average moulded fresnel lens. The old Luxeon 1 always needed a small secondary lens to increase its virtual source size sufficiently to match the fuzzy focal patch of the fresnel. The new lambertian Luxeon III eliminates that need and simplifies the transmitter's optical design. Even with moderate increases of source intensity, other optical advantages of the new red Luxeon III are valuable in light beam communication. The increased source size provides a greater dispersal of the transmitted beam, which permits simplified transmitter mounting and aiming arrangements. A major disadvantage of laser communication is its need for massive rigid mountings with directional micro-adjustment in altitude and azimuth. Arrangements of this type really are impractical in the portable equipment necessary for field usage on mountain tops or in cars, particularly in a windy environment. By contrast, our own mountings are standard foldable photographic tripods, or, in Mike VK7MJ's case, a portable 'Black and Decker' saw-bench provided with wooden wedges to tilt the fresnel box! Our aspiration has always been maximum sig/noise for minimum cost, which we feel is the only sensible and practical approach to the problem.

At a price of around Aust.$29, with easy availability of the product throughout Australasia through the efficient mail-order service of Techbits, and with a life of around 20,000 hours at 1400mA drive, the red Luxeon III (part number LXHL-LD3C) represents good value. The price of these Luxeon devices is slowly dropping with time...

We are indebted to Lumileds Incorporated for most of the graphs and diagrams included in this web page, and our own measurements indicate that their light output measurements tend to be slightly on the conservative side of the mean value!


Indicator-type LED's, commercially produced with Gallium Arsenide Phosphide chips since 1968, were always limited to a dissipation of 100mW or less by their design. Their epoxy encapsulant couldn't withstand temperatures above 120 C, and the wire leads couldn't dissipate the excess heat generated by the chip:

Commencing in 1998, Philips Industries joined the former Opto-Electronics Division of Hewlett-Packard, by then known as Agilent Technologies, to form a new business, Lumileds Incorporated. Lumileds developed the 'Power LED' concept, re-designing the way in which the diode generated light and dissipated heat. The Luxeon's most obvious improvements are the diode chip's immersion in a transparent silicone gel with greater heat resistance than epoxy, and the thermal bonding of the chip to a thick copper or aluminium heatsink slug. In previous incarnations of the device, this thermal bonding was achieved with electrical isolation, but the new Luxeon III's slug is welded to its metal-backed PC board of about 2 cm in diameter in the now-familiar brooch-like 'Luxeon Star' configuration - see the picture at the top of this page. The interior workings of a typical Luxeon are shown below in cross-section - in this case using the InGaN chip to generate green, blue and white light. The device is nearly always sold pre-mounted to its metal-backed star shaped PC board, for ease of coupling to a heatsink:

Lumileds' effort has principally been directed to making an efficient white illumination source to compete with incandescents. For this, they use a blue-emitting InGaN chip is with a yellow phosphor coating. Naturally, the response of a phosphor-based light is much slower than the chip's direct emission, so these are unsuited to communications. The short wavelength blue-white radiation also doesn't match the spectral response of silicon photodetectors. The full range of Luxeon colours available is shown below:

Of the available Luxeon emission colours shown above, the red, red/orange and amber emitters have the advantage of being based on the AlInGaP chip, which can be driven to a much higher current density than the Nitride chips which generate the shorter wavelengths. The blue-cyan-green-white Nitride chip can only reach 50 amps/cm2, while the red-orange-amber AlInGaP chip can be driven at 140 amps/cm2. Red Luxeons also provide the best spectral match for silicon photodiodes, as the response graphs below indicate:

Fortunately, the niche market for red sources in traffic control signals and in automotive tail light indicators has induced Lumileds to manufacture them in quantity. These 630 nm Luxeons are based on the exceptionally high efficiency AlInGaP chip. The rise time of this chip is of the order of 15 to 25 nsec. The construction of the red AlInGaP Luxeon differs from the blue-green-white emission InGaN-based structure in several respects, as we see below:

A unique feature of the red AlInGaP Luxeon is its usage of the 'truncated inverted pyramid' ('T.I.P.') chip configuration developed by Mike Krames and his colleagues at Hewlett Packard in 1994. Conventional LEDs had several internal loss mechanisms, the principal one being a result of their chip materials having a high index of refraction. Only 30% of the photons generated could be extracted from the standard rectangular chips, partly as a result of total internal reflection at the chip's surfaces. By cutting the chips into an inverted pyramid shape, its walls reflect the photons forward, or pass them to the surrounding metal cup, which reflects them forward. This improves the proportion of emitted photons from 30% to 55%.

The red Luxeon 1 (1 watt input) incorporating these and other improvements, was released commercially around the year 2001. The red colour was not available in higher power until April 2005, when it was offered in the Luxeon III (4.6 watt max.) format. The Luxeon 1 was the basis of our record-distance atmospheric communications contacts in 2002 and February 2005, and its performance was excellent. However, the ruggedised and slightly enlarged red Luxeon III makes its Luxeon 1 predecessor appear puny by comparison, as the following graphs of power input and light output  indicate:

A device with the red Luxeon III's power rating naturally produces some loss via heat. The importance of dissipating that heat is shown by the graph below. The light flux output of the red Luxeon III is inversely related to the temperature at which it operates. The larger the heatsink on the Luxeon III, the greater its light output can be:

The amount of heatsinking required for safe operation of the red Luxeon III is shown below. It's interesting to note that this graph implies a maximum continuous current drive of 1540 mA, while some published specifications recommend only 1400 mA. In spite of the 'Luxeon III' name with the implication of a 3 watt rating, its maximum would appear to be around 4.6 to 4.8 watts! These devices demand much larger heatsinks than their 1-watt predecessor:

For coupling into a fresnel collimator with an f/d ratio approaching unity (ie lens diameter ≈ lens focal length), the 'lambertian' dispersal pattern of the Luxeon III provides a near-ideal match. We've taken the original graph of this dispersal from the literature published by Lumileds, and have re-drawn it onto polar co-ordinates to provide this better visualisation of its optical performance:

While the 'lambertian' radiation pattern suits our purposes perfectly, our readers should be made aware that Luxeons can also be supplied with 'batwing' or 'side-emitting' polar patterns. These are unsuited to collimation and should be avoided for communication purposes.


Modulating the power input to the Luxeon III is a fairly straightforward process at audio baseband frequencies.

Atmospheric amplitude scintillation of the light beam is the greatest problem on long-distance optical links. The usage of an FM or digital subcarrier to convey audio would seem advantageous. However, our field trials with these rather complex subcarrier systems indicate that they give poor results when the light signals are very low. Simple amplitude audio modulation of the beam will always win at low signal levels, because:

Much greater bandwidth is needed for an FM subcarrier and its sidebands than for baseband audio. Broader bandwidth always involves a higher received noise level. The optical receiver noise power per Hz of bandwidth (ΔHz) is fairly constant across the audio spectrum. The FM signal has to be much further above this noise level than a simple analogue baseband signal to reliably reconstruct the high-frequency subcarrier at the receiver with a Schmitt/clipper, particularly when scintillation is occurring.

Inexpensive P-I-N photodiodes cannot simultaneously be optimised for high sensitivity and fast response speed. As the value of the feedback resistor in a transimpedance-configuration photodiode preamplifier increases, the PD-amp combination increases in sensitivity to light (ie signal-to-noise ratio), but the bandwidth drops.

Owing to these two processes, optical DX signals received at very low sig/noise levels will be at a distinct disadvantage in employing an FM subcarrier. In marginal signal DX work, beam scintillation  is better alleviated by alternative means, for example:

(1) The usage of very large transmitting and receiving apertures, which average the signal levels passed via adjacent atmospheric turbulence cells. In practice, 40 cm by 60 cm fresnels often straddle several adjacent optical turbulence cells, which are commonly 7 to 20 cm in diameter. Large optical apertures also greatly increase the signal gain of the system. These large apertures are equally important at the transmitter and the receiver.

(2) The usage of spatial diversity reception, with multiple optical receivers spaced 2 or 3 metres apart, so that fading nulls do not occur simultaneously at all receivers. This technique was advantageously employed on our 167.7 km contact at the South Mount Barrow Peak, Tasmania (Australia) on 19 February 2005.

(3) The usage of narrow bandwidth tone modulation telegraph systems, the so-called 'digital' communication modes, such as JT44 and JT65. Many of these have been designed to work in the presence of scintillation.


For audio transmission, one needs a driver to apply a DC bias to the Luxeon III, and steer the current equally upwards and downwards to apply the modulation. For simple analogue amplitude modulation, the Luxeon III's standing current should be set to around half the device's permitted maximum, around 700 mA, with a series resistor limiting current peaks to just a bit more than the maximum continuous rating, say 1.6 Amps. The power supply for this arrangement would need to have a very low impedance. A 10,000 F power supply reservoir capacitor should provide adequate peak current capability with reasonable ripple current reserve. Dry primary ('torch') cells, even of 'D' size, would be inadequate. No battery supply smaller than about a 4 Amp/hr gel-cell would be suitable, and for full modulation with the 3.5 volts peak voltage drop of the Luxeon III, the rail would have to be 9 volts or more. To make the transportable modulator compatible with automotive supplies, a 12 volt rail would be a good choice, though only around 25 % of the power would result in useful input to the Luxeon III. The remainder would be dissipated as heat in the modulation transistor, and in the Luxeon III's current-limiting series resistor.

With an output transistor running in grounded emitter class A, the Luxeon III could be placed between the collector and rail with its current limiting resistor in series with it, as shown below:

Alternatively, the output transistor could run in emitter follower configuration with the Luxeon III and limiting resistor between emitter and earth. With a 12 volt rail, if the output transistor had a typical beta of 20, and the Luxeon was passing 700 mA, the base would have to bleed about 700 20 = 35 mA, and a 700 mW driving signal should be more than adequate. This audio drive could be inexpensively provided with a minimum component count by an LM386N IC (commonly available in Australia).

The LM386 has an output pin that sits automatically at half rail voltage - around 6 volts. Some adjustment of the output transistor base's operating voltage through series diodes (each dropping 0.7 Volt) will ensure the correct quiescent current conditions through the Luxeon.

We should emphasise that the circuit diagrams above are only suggestions for prototyping, and haven't yet been tested. Our present modulators are set up for the Luxeon I, which needs a standing current of only 190 mA. To produce the Luxeon III's requisite quiescent current of about 700 mA, the voltages around the output transistor would probably have to be manipulated and changed from the values shown. These are 'ball park' component values, for experimental circuit development only.


A suggestion made by Burlinson and Averay in the Australian 'EEB' magazine circa 1970 was to use pulse width modulation for modulated light transmitting. A sawtooth waveform may be added to an audio waveform, then applied to a Schmitt trigger to produce audio-modulated PWM, as shown in the diagram below.

If the light signal is very low and close to the noise on receive, a slow and sensitive photodetector would receive the PWM signal in analogue mode, simply reproducing the average (audio) light level as a result of its slow response. With higher received light levels, a faster photodetector could clip, limit and reconstitute the transmitted PWM signal through a Schmitt trigger, eliminating atmospheric amplitude scintillation. In this way, a PWM signal would be advantageously compatible with either analogue or digital demodulation techniques.

PWM would eliminate the need for linearity in the emitting diode and in the modulating transistor, which could both operate in switch mode. The switching transistor modulator could achieve a much higher efficiency than a linear modulator, dissipating virtually no heat. It would also permit the modulator to work from a lower rail voltage, say 6 volts instead of the usual 12, with additional efficiency advantage.


Lumileds Technical Datasheet DS46: 'Power Light Source - Luxeon III Star' pub.: Lumileds Lighting, San Jose, USA, 2005.

Lumileds Publicity Release PR35: 'Lumileds Shatter Performance Records with Release of 190 Lumen LUXEON III', [on the release of the red, orange and amber Luxeon III], pub: Lumileds Lighting, San Jose, USA, 31 March 2005.

Lumileds Application Brief AB25: 'Luxeon Reliability', pub: Lumileds Lighting, San Jose, USA, February 2004. Contains a detailed explanation of the internal differences between the various Luxeon chip designs.

Steigerwald, Bhat, Collins, Fletcher, Holcomb, Ludowise, Martin & Rudas: 'Illumination with Solid State Lighting Technology': IEEE Journal on Selected Topics in Quantum Electronics, Vol. 8, No. 2, March/April 2002 pps. 310 - 320.

Schubert, Fred E: 'Light Emitting Diodes', Cambridge University Press, UK, 2003.

M R Krames et al: 'High Power Truncated-Inverted-Pyramid (AlxGa1-x)0.5In0.5P/GaP Light Emitting Diodes Exhibiting> 50% External Quantum Efficiency', in Appl. Phys. Lett., Vol 75, pps 2365 - 2367, 1999.

Bergh, A A & Dean, P J: 'Light-Emitting Diodes', Proceedings of the IEEE, Vol. 60, No. 2, February 1972 pps 156 - 224. An expanded book was published by these authors under the same title by Oxford University Press (UK), in 1974 and Clarendon Press (USA), in 1976, 591 pps.

Neuse, C J; Kressel, H & Ladany, I: 'Solid State: The Future for LEDs' in IEEE Spectrum, Vol. 9, No. 5: May 1972.

'Scientific American', February 2001 - article on power LEDs (details to be posted - not to hand at the time of writing this web page!)

Web page: On truncated inverted pyramid (TIP chip) LEDs.

See also (via Lumileds site): 'IESNA National Conference 2004 LED Paper'

Back to Modulated Light DX

Last update to this page: Sunday 29 Jan, 2006 (fix Luxeon link that had moved)

Page count since 201005: