Optical communications
using coherent and
non-coherent light

An overview

Left:  A 3-watt red Luxeon LED at a distance of 14.91 miles (23.85 km) with downtown Salt Lake City in the foreground.
Right:  Transmitting with a 50+ watt LED with the light from the 95 mile (152km) distant end being visible at the terminus of the red beam.
Click on an image for a larger version.

Luxeon 3 watt LED at a distance of 15
                            miles; Downtown Salt Lake City is in the
                            foreground The south end of a 95 mile (152km)
                            optical path near Salt Lake City

Other pages here at modulatedlight.org

  • Operation Red Line 50th Anniversary! - The historic May, 1963, 118+ mile optical transmission by the EOS Amateur Radio club.  The remarkable feat was accomplished just months after the invention of the visible-light HeNe laser!  
    • Operation Red Line photo gallery - Pictures and detailed descriptions of the events and equipment used during the May, 1963 experiment.  This and the above page was produced with help of some of the participants of that historical event.

Unlike more familiar fiber-based schemes, communications through the atmosphere has some unique challenges that do not confront those who might use optical fiber links such as distortion caused by variations in the air itself along with weather and pollution that can diminish the signals - not to mention both artificial light and the sun which can overwhelm distant, weak signals!

Note that the emphasis here is on experimental schemes and those described here are not intended to be either highly reliable or high-bandwidth.  Because of this, the commercial viability of some of the techniques described here is likely irrelevant:  Much of the goal of this experimentation is to simply try different things and to have fun while doing it - and hopefully learning something that we didn't previously know in the process!

Coherent versus non-coherent light for through-the-air optical communications

"To lase, or not to lase..."

have been the darling of optical through-the-air communications for decades - and for some fairly good reasons:
Historically speaking, Lasers are relative newcomers in the optical communications field:  For many centuries, reflected sunlight or flames have been used to convey messages visually over long distance while more recent electronic schemes have used electric lights of some type (incandescent, gas discharge) to provide a source of modulated light - but high modulation depth with good frequency response had long been a problem.  It turns out to be very difficult to modulate most "conventional" high-intensity, non coherent light sources with properties of both full modulation and/or usable frequency response.  While there are various schemes that can accomplish this, they often are quite complicated and/or beyond the means of the average experimenter.

Lasers, on the other hand, would seem to lend themselves quite nicely to long-distance optical communications with gas lasers and readily-available laser diode assemblies already designed and packaged to produce collimated, intense light.  It is quite easy to electronically modulate a laser diode, and either gas or solid state lasers can be modulated mechanically at very high frequencies with an optical cell.  One sticking point with most lasers is that it can be quite a challenge to linearly modulate the laser itself electronically over a wide dynamic range so it is more common to use schemes such as an FM subcarrier or digital modulation (usually pulse-width schemes) that can be accomplished simply by turning the light on and off rather than trying to change the brightness in some linear way.

More recently, extremely intense, non-coherent solid state light sources have appeared in the form of high-power LEDs:  These devices have luminous outputs higher than that of most lasers - at least ones that an experimenter is likely to find or afford!  With these recent technology advances it is practical to get fairly high conversion efficiencies from single-color, high-power LEDs, easily achieving hundreds of milliwatts (or even watts!) of light output with manageable power input levels.

Like traditional LEDs, high-power LEDs lend themselves nicely to current modulation and their upper modulation frequencies are limited, for the most part, by their intrinsic capacitance, making them practical even for video or megabit-rate data modulation.  The downside is that these LEDs do not usually come in a form that produces tightly-collimated light so it is up to the user to add the optics necessary to make them useful for use in a long-distance communications system.  Another important consideration is that, unlike lasers, LEDs do not produce coherent light - but this can be a distinct advantage as we shall soon see!


Another consideration when using lasers has to do with safety.  Compared to even a low-power (e.g. Class 2) laser, the highest-power LED-based systems have lower power density across the area of collimation.  Much of this has to do with the fact that, by necessity, it is necessary to collimate the beam of an LED emitter to a fairly large area (typically hundreds of square centimeters) to minimize divergence - a factor that greatly reduces the power density - and it is for this reason that practical, high-power LED-based systems are a minimal physical hazard to those using them or to those who might inadvertently cross into the beam at a distance and are thus unlikely to be regulated in the same way as lasers.

It should be noted that in the U.S., higher-power lasers (those that exceed Class 3R/3A levels) are restricted in their use except in controlled environments and/or unless a proper variance has been obtained - and similar laws (often ones that are even more restrictive) are commonly found in other countries as well.  Finally, in some jurisdictions - such as New York, Arizona and Texas, lasers may be subject to stricter "local" laws while in places like New South Wales, Australia, laser pointers have been effectively been banned altogether!

If you really want to try your hand using laser pointers for communications, see the "Using Laser Pointers..." page at this site.

For a historical overview of optical communications see:

Large, collimated beams are best:

Whether one is using a Laser or not, successful long-distance optical communications requires that the emitted beam be collimated - preferably into as large a diameter, parallel beam as possible.  While it may intuitively seem like a good idea to preserve the small-diameter, extremely intense, pencil-like beam of a typical laser, it is important to realize that this beam has a very small cross-sectional area.  Because the atmosphere is turbulent by nature, as the beam passes through small "cells" of air with different characteristics (temperature, humidity, pressure) the path of the light beam is refracted (bent) slightly every time it does so.  A very small-diameter beam of light will, therefore, slice its way through a narrow path of air cells, accumulating more and more disruption as it passes along with the result that, at the far end, the received beam may have traversed enough of these cells of air to be significantly disrupted and randomly vary in brightness by the time it reaches the receiving end - a phenomenon known as scintillation .

A large, collimated beam, on the other hand, has lower power density and may appear dimmer to the naked eye to an observer relatively near the emitter, but only because the energy has been spread out over a larger area as compared to the pupil of the eye.  Because of its larger area it is more likely to be able to "straddle" multiple cells of air - or even small visual obstructions like rain drops or insects:  While portions of this larger beam's cross-section may be degraded by air cells, other portions are likely to not have been so-effected at the same instant and because of this, the overall amount of scintillation is reduced - an effect called aperture averaging.

In part, this effect can be easily demonstrated in the night sky:  Usually, stars twinkle, but planets don't!  A star is, for all practical purposes to the naked eye, a point-source of light because, while the star itself may be large, the vast distances in space make its subtended angle negligible in size.  A planet, on the other hand, being very much closer, is not simply a point of light:  Even a small telescope will resolve the nearby planets (Venus, Mars, Jupiter and Saturn, in particular) as obvious disks rather than points of light and in this way it's easier to see how aperture averaging of the planet's disk would reduce twinkling:  Even if one portion of a planet's disk "twinkles out" it's likely that another portion of the disk is still shining bright at that instant!

Another contributor of the "stars twinkle, planets don't" phenomenon is a property often referred to as "local coherence" which is at play when the angular source size is very small.  This effect, noted by A.A. Michelson, is another of the causes of scintillation.  If the emission of light is from a source with extremely small apparent angle (like the "pinpoint" of a star) even noncoherent light can take on many of the properties of coherent light - notably interference.  For more information on this topic read "The Sizes of Stars" by Calvert, and for a graphic example of this effect read the Astronomy Picture of the Day for April 28, 2011.  This effect is yet another reason why a large aperture will contribute toward the reduction of the phenomenon of local coherence and thus scintillation.
Figure 1
The Top trace, covering a time span of about 0.8 seconds, shows 4 kHz signal emitted by a standard, high-brightness red LED being received over a 15 mile (24km) path having been emitted from large-area (approx. 50 sq. in, or about 289 sq cm) aperture.  This shows about 17dB of scintillation at a relatively slow rate.
The Bottom trace, covering a time span of about 0.28 seconds, shows 4 kHz laser signal, over the same path, using the very same optics as above for both receive and transmit.  Close inspection of this waveform reveals at least 40dB of scintillation occurring at a much higher rate than with the LED.  The signals were both received using the same 70 sq in (452 sq cm) Fresnel lens optical receiver.
Click on either image for a larger version.
                    of a signal emitted from collimated lens
                    of signal emitted from the same aperture as the LED

It should also be remembered that the amount of scintillation also depends on the size of the aperture being used as an optical receiver:  The iris of the eye, being only millimeters in diameter, is very small when compared with the objective lens of even a small telescope.  For this reason, stars that appear to twinkle to the naked eye often appear to be far less "twinkly" when viewed with a telescope or binoculars.  It is also worth reminding the reader that with a larger "receive" aperture (e.g. larger lens) more light is gathered and the effective sensitivity of the receiver is increased in proportion to its increase area!

Scintillation and coherent light:

One unique property of lasers is that they produce light that is (more or less) emitted at a single frequency or wavelength and ideally, the light leaves the laser in phase-coherent wavefronts.  If one were to cause a portion of the light beam the light to be delayed slightly as compared with another portion, wavefront cancellation can occur - and this is precisely what happens when an interference pattern is generated:  This phenomenon can be readily demonstrated using a CD or DVD to reflect laser light and noting the resulting pattern of dots being reflected, or it may be inferred by that familiar "laser speckle" that one sees on a surface illuminated by laser light.

This very property has a downside in through-the-air communications:  Parcels of air of varying density due to temperature and/or humidity can offer slightly different velocities of propagation to the light in addition to slight refraction.  The ultimate result of this is that scintillation of a coherent light beam is usually very much worse than a non coherent beam due to diffraction, owing to the fact that random wavefront phase reinforcement and cancellation is occurring.

Figures 1 and 2 illustrate this point clearly.  For Figure 1, the same 15 mile (24km) path was used, along with the same detector optics (70 sq in, or 452 sq cm Fresnel lenses) and the same 50.27 square inch (289 sq. cm) optics (a reflector telescope) being used for transmitting.  The top image shows the amplitude of the LED (noncoherent light) being affected by the scintillation to a depth of about 17dB or so.  The bottom trace shows the received Laser signal being much more strongly affected - and at a faster rate.  Note that the two images in Figure 1 have different horizontal time scales with the bottom image representing a much shorter time period and far more severe scintillation.

In the video linked in Figure 2 the differences are rather dramatic.  Comparing the "raw" small-diameter beam from a cheap, red Class II laser pointer with that of a collimated beam from an LED the results are visually striking:  As noted in the narration the "speckle" pattern from the distant laser is a chaos of light while the effects on the light from the LED - while significant - are of far lower magnitude.  From this video one can see that the use of both coherent light and a large transmit aperture can be helpful in maintaining the integrity of the signal as it passes through the atmosphere.

For an audio clip that demonstrates the difference between Coherent and Noncoherent light and their passage through the atmosphere, click here.

Another property of coherent light that should not be overlooked is that due to absorption by various gasses, the transmissivity of the atmosphere has many narrow peaks and nulls throughout the visible spectrum and a "narrowband" light source (such as a laser) could easily fall into one of those nulls:  Noncoherent light sources, by their very nature, are not likely to be as susceptible to the effects of very narrow nulls in the transmissivity although from a purely practical standpoint, inexpensive diode lasers have wide enough spectral width that these extremely narrow nulls are largely "straddled" in normal operation.

Comments about lenses, coherent and non-coherent light emitters:

Again, no matter what light source is to be used it is advantageous to use as large an optical aperture (lens) as possible in order to minimize scintillation - not to mention to maximize "optical gain" at the receive end.  Furthermore, when laser (coherent) light is used, best efficiency is obtained when the quality and precision assures that these optical components are "diffraction-limited" - that is, made such that their figure is accurate to sub-wavelength accuracy so that it is the finite wavelength of the light itself that is the main determining factor!  If one keeps this consideration in mind, there are a few factors that must be considered when obtaining/using such optics:
When one is using a laser to provide a coherent source for a collimated beam it makes sense to use good-quality optics to minimize beam divergence, but this poses a number of problems (in addition to cost) that severely limits the size of lens that one is likely to be able to use - not to mention being able to transport the optics and associated mounting gear.  These complications aside, the optical alignment (e.g. aiming) of a very tightly collimated beam over large distances requires extreme precision and mechanical stability - both factors that complicate the logistics of experimentation while in the field.

A practical alternative to a large glass lens is a plastic, molded Fresnel lens.  These lenses are flat and, if properly manufactured, can have excellent optical characteristics - including absence of spherical aberration.  Affordable plastic Fresnel lenses are, however, no match for a good quality set of glass (or optical plastic) "conventional" lenses in terms of performance (e.g. they cannot approach the diffraction limit at visible-light wavelengths) and because of this they cannot be used to efficiently collimate a coherent light source such as a laser.  Even though Fresnel lenses are not suitable for collimating coherent light owing to their rather poor accuracy (typically, hundreds of times worse than the diffraction limit) they may still be used at the distant "receiving" end because of the fact that the atmospheric path will "de-cohere" the light, anyway!  Another downside of Fresnels is that the "facets" (grooves) the lens tend to scatter light which can be an issue if there are nearby light sources that may be strong enough that the scattered light might "dilute" and interfere with the desired signal, but this problem can be significantly controlled through the use of light-blocking baffles and careful site selection.

Inexpensive Fresnel lenses can provide excellent results as a beam collimator when using noncoherent light sources such as high-power LEDs but they are of limited usefulness if one is trying to collimate coherent (e.g. laser) light owing to their innacuracies.  When using good-quality Fresnel lenses with typical high-power LEDs it has been observed that it is the size of light source itself (e.g. the LED's die) and not the quality of the lens has been the main factor in determining beam divergence and values of well under 0.3 degrees (approximately 5.2 milliradians) are easily attainable with readily-available light sources - and values significantly lower than this would be possible if a high-intensity "point-source" LED was practical:  Unfortunately, such a "point-source" LED can not exist!

Even though the use of high-power LEDs with their relatively large emitter areas implies a more-divergent beam than that obtained from a typical laser, the sheer magnitude of luminous flux available from the LED still allows a combination of respectable far-field flux, larger transmitting aperture, and much less-stringent requirements in aiming - all of which, in practical terms, allow for excellent in-field performance.  Another important consideration is that in using Fresnel lenses, lenses large enough (>25-30cm equivalent diameter or larger for visible wavelengths over a 100km path) to minimize the effects of the "local coherence" and disruption due to the "cells" of air described above are quite affordable and practical in their use!
Figure 2
A video showing the effects of scintillation on small-diameter beam from a laser pointer and that from a collimated LED at a distance of approximately 24km.

In a receiver, Fresnels are very effective at intercepting the distant light source and the ability to inexpensively achieve large aperture areas can make up for optical inefficiencies that they might have in comparison to smaller-sized, high-quality lenses.  As is the case with the use as a collimator, the relative imprecision of a typical molded Fresnel lens sets a limit as to how small a "spot size" can be achieved - a factor that has implications related to small the detector can be as well as the minimum practical field-of-view.  Over atmospheric paths one may use a Fresnel to intercept light from a distant coherent source without the same diffraction problems that one would encounter were it used as a coherent light collimator, this being because of the fact that even a short distance through the air (a few kilometers) will adequately "de-cohere" the laser's light.

Methods of signaling:

Perhaps the simplest form of signaling is simple on/off keying of the light, but this scheme is not well-suited for electronic detection of weak signals so Morse code is often used with a tone-modulated light source, being detected by ear with an optical receiver at the far end.  This scheme has the advantage that it moves the detected signals into the realm of one's "Gray Matter DSP" (that is, the brain) via the ear and a skilled operator can easily "copy" signals that are buried in noise.  On/off tone modulation is also fairly easy to accomplish:  Simply interrupt a tone-modulated light source (or the tone itself) to send the Morse characters.

Tone modulation may be done in a number of ways and one of the oldest is the "chopper modulator."  Used by Alexander Graham Bell in his early Photophone experiments, this device simply interrupts the light source - usually with a spinning, slotted disk to impose audio onto the light source and this modulated light source is, itself, interrupted to "key" the tone.  This mechanical scheme has the advantage of being intuitively obvious and it may be used to modulate practically any light source!  Alternatively, tones may be imposed on the light and data communications may also occur with very slow data rates, using a computer at each end, allowing information to be conveyed even if the signal is too weak to be audible to the ear.

Amplitude modulation of plain speech is highly attractive in that it does not require that the operators be skilled in Morse in order to communicate, but the use of speech can complicate things as it is difficult to satisfactorily modulate it onto many light sources - such as a slow-to-respond tungsten filament.  It is possible to modulate other high-intensity light sources such as arc or gas-discharge lamps provided that one deal with the complexity of dealing with the awkward voltage and/or current requirements of such devices and accept that fact that the depth and frequency response may be limited by the nature of the device.  Direct speech modulation of gas lasers can be complicated as well, owing to the nature of many laser tubes to resist modulation to a significant depth.  Both gas and semiconductor lasers can be modulated with an optical cell (such as a Kerr cell) but these pose their own complications, such as the need for a high-voltage source and, possibly, the use of dangerous substances.

Modern laser diodes may be directly current-modulated provided that one strictly observes the ratings pertaining to minimum and maximum current and device temperature whereas LEDs are far more-easily current-modulated.  As noted previously, it is more common that laser diodes are simply on/off modulated - either with an audible tone to facilitate Morse communications or with a much higher frequency where Frequency Modulation or FM may be used - that is, the Laser (or LED) is simply turned on and off, but the varying frequency is used to convey the modulation be it voice, video or even data:  More on these and other methods later.  For "linear" modulation schemes, duty-cycle modulation (e.g. PWM, or Pulse Width Modulation) is a relatively simple method where simple on/off modulation can be varied in a way that it will synthesize linear modulation with the advantage of not having to worry much about the actual linearity of the device being modulated - such as with a laser diode!

Types of detectors:

For reasons of practicality, most systems for detecting optical energy - including those that we have used - have been radiometric - that is, we are simply detecting energy from the transmitter in a manner that is intrinsically frequency or wavelength insensitive:  The more light we receive from the distant transmitter, the more signal we can recover from our detectors.

Perhaps the most inexpensive type of detector is the silicon PIN photodiode.  These devices are particularly sensitive in the red and near-infrared spectrum - which just happens to be about the same as the optimal wavelengths for through-the-air communications.  The problem with photodiodes is that in order for them to be very sensitive they need to be very lightly-loaded owing to their lack of any intrinsic self-amplification - but if you do load them lightly enough to get good sensitivity, their capacitance (10's to 100's of pF for units of several square millimeters in area) can seriously limit high frequency response:  If one wishes to obtain the ultimate in sensitivity from a photodiode, best sensitivity is only possible up to a few kHz with the optimum range being below a few hundred Hz.  These frequency limitations effectively rule out using any subcarrier scheme to convey voice or high-speed data if one wishes to have, simultaneously, both wide frequency response the and best-possible sensitivity.  For these reasons our experiments have tended to use simple amplitude-modulated voice as well as extremely narrowband digital techniques such as WOLF, WSJT or QRSS (extremely slow Morse) - all in or below the 3kHz speech range.

An alternative to using PIN photodiodes is to use Photomultiplier tubes (PMT's).  Photomultiplier tubes have the advantage in that they can be extremely sensitive owing to their self-amplification properties while maintaining excellent bandwidth - but they do have some disadvantages:  They are rather expensive, especially compared with a photodiode,  they require a high voltage supply - 1000 volts being typical, and most commonly-available types have rather poor red-wavelength sensitivity, an important factor when you consider that these longer wavelengths are preferred for through-the-atmosphere communications.  Other disadvantages of PMTs is that in comparison to a photodiode, they are rather fragile, both mechanically and electrically.  A solid-state alternative to the PMT is the Avalanche PhotoDiode (APD) as these devices can have excellent red sensitivity - greatly exceeding that of many surplus PMTs - but APDs, like PMTs, tend to be somewhat specialized and expensive and low-cost devices are not always readily available on the surplus market..

For more background on detectors, see the page "Optical Receivers for low-bandwidth through-the-air communications" and its related links.

Why not FM subcarriers?

Many previously-published articles have used FM in the form of subcarriers to convey voice information - and for some technically-sound reasons:
There is a problem that can arise when trying to demodulate signals that are near the noise threshold of the typical PIN-diode optical detector system:  When using photodiodes the available sensitivity decreases with increasing frequency response.  What this means is that a receiver that will receive, say, a 40 kHz subcarrier, will likely be at least 10-20dB less sensitive than a receiver optimized to receive only speech (up to 3 kHz) bandwidth.  Another factor has to do with the fact that a skilled listener can comfortably copy speech with only an 8-10dB signal-noise ratio - and this happens to be approximately the same amount of signal-noise ratio that is required for an FM demodulator to work.  If you have plenty of excess link margin, however, a subcarrier-based system can work quite well.

How about other types of subcarriers?

Experimentation has been done by the folks in Great Britain in 2010-2011 with the use of VLF transverters to convert signals from an HF amateur radio rig (such as an FT-817) to/from frequencies typically around 3.58 MHz from/to "subcarrier" frequencies in the 25 kHz range that are to be used with the optical gear.  While the use of these higher-frequency subcarriers will reduce the effective sensitivity of most types of optical detectors, the using of SSB in favor of FM minimizes detection bandwidth and the selection of the lowest-possible "carrier" frequency (e.g. 25 kHz or lower) can be used to further minimize the effects of the loss of sensitivity at these higher frequencies.

It is worth remembering that the goal here is not to obtain absolute maximum performance, but to shift the audio up and away from the "hum and buzz" that might be encountered from the effects of mains-powered urban lighting.  Such a scheme also makes use of an "IF Rig" that many VHF/UHF/Microwave enthusiasts (some of whom may be interested in optical communications!) would already have, allowing the use of any mode provided by the radio (such as FM and SSB) - as long as the signal is good enough!  The downside is that not only must one tack on a VLF transverter to a modulator/detector that already provides audio, but you must supply an amateur HF rig and the power to run it as well!  If you already have such gear - and your goal isn't intended to be that of achieving ultimate DX (that is, you just want to work other stations a few 10's of km away rather than 100's of km) - then this may be a reasonable alternative.

In lieu of using a transverter, I have also used the "Spectrum Lab" program with a laptop computer to produce and demodulate SSB, AM and FM signals in the 5-24 kHz frequency range and have produced a number of rudimentary "scripts" - all of which could stand further development, but have adequately demonstrated the feasibility of doing so using computer sound cards.

(If you are interested in methods of communications using subcarriers, feel free to use the contact information at the bottom of this page.)

Final comments:

It seems fairly clear that the majority of the research in through-the-air optical communications has been directed toward short range (under just a few kilometers) use - and for good reason:  The variability of the atmospheric conditions (not to mention daylight) simply prohibits the use of such techniques for use as a full-time, highly-reliable communications system.  As mentioned above, the goal is not to attempt to create an ultra-reliable, high-speed, optical through-the-air communications system, but rather see what we can do with fairly simple and inexpensive hardware.

If you have questions or comments concerning the contents of this page, feel free to contact me using the information at this URL.

Return to the main "Optical Communications" web page, or go to the modulatedlight.org main page, or
go to the ka7oei.com  page.

Keywords: Keywords: Lightbeam communications, light beam, lightbeam, laser beam, modulated light, optical communications, through-the-air optical communications, FSO communications, Free-Space Optical communications, lightbeam communicator, LED communications, laser communications, LED, laser, light-emitting diode, lens, Fresnel, Fresnel lens, photodiode, photomultiplier, PMT, phototransistor, laser tube, laser diode, high power LED, luxeon, cree, phlatlight, lumileds, modulator, detector

This page and contents copyright 2007-2013 by Clint Turner, KA7OEI.  Last update:  20130726
Count since July, 2013: