OPTICAL COMMUNICATION FOR THE AMATEUR

by Chris Long, originally published in 'Amateur Radio' magazine by the Wireless Institute of Australia, Melbourne, January 1979, pps 7 - 14.

My article, 'Optical Communication for the Amateur', published 26 years ago in 'Amateur Radio', was to my knowledge the first serious foray of Australia's official ham radio publication into the realms of atmospheric optical communication.

In 1979, at the age of 24, my grammar was a little clumsy and my optical knowledge was rather sketchy, so I hope that I've rectified this in the current web page edition! This data was presented when all laser sources were prohibitively costly. LED's were then pathetically limited in their output flux. According to Hewlett-Packard's 1976 "Optoelectronics Designer's Catalog" (p.5), SEVEN (7) milliCandela at 30 mA was then considered "high efficiency". Compare this with the red 'Luxeon III' LED released in March 2005, producing FOUR HUNDRED THOUSAND (400,000) mCd at 1,400 mA. Understandably, in 1976 we had recourse to gas discharge lamps and high pressure mercury arcs as modulated sources. While Luxeons and diode lasers are more convenient and efficient for most atmospheric optical comms systems today, the Hg arc lamp is still a viable source for experiments with 'cloudbounce' and non-line-of-sight (NLOS) operation where narrow transmitted beamwidths are of little consequence. In this case, where reception will be via diffuse specular reflection, the total amount of flux reflected or scattered by the cloud is more important than the breadth of the beam illuminating that cloud. In this instance, 'brute forcing' the system by using a mercury arc in the prime focus of a searchlight reflector may prove more practical as a means of recovering dispersed flux than using a finely collimated gas laser with a power output only in the order of a few milliwatts.

The use of audio-modulated light beams for communication pre-dates the first radiotelephone experiments by almost 25 years. In 1880, Alexander Graham Bell and Charles Sumner-Tainter used vibrating mirror systems to superimpose sound modulation on reflected beams of sunlight.

With receivers employing selenium photoconductive cells, ranges of about 700 feet were spanned by this 'photophone' system. With improved mechanical light modulators and an electric light source, A O Rankine demonstrated a system with a range of several miles in 1916.

From time to time over the last eighty years, enthusiasts' radio magazines have published articles on modulated light communications, though few have involved equipment capable of ranges comparable with radio. A typical example from 'Wireless World' in 1930 is shown below:

The German, Japanese and Australian armies did some of the first communication experiments with modulated electric light sources around 1935-40, using techniques derived from the recording of optical sound tracks on motion picture film. It is believed that some of the first research done by the same Mr Sony who established the Sony Corporation in Japan involved prototyping a modulated light communication system in the early 1930s. The high directivity and security of these systems gave them obvious military applications before microwave hardware became available.

Australia's contribution to modulated light research in wartime was of a limited nature. Under the direction of a Colonel Johnson, Corporal Bob Slutzkin (VK3SK) of the Australian Army HQ Signals Group developed vibrating reed light interrupters, modulated incandescent filaments and related optical systems during the early part of the war, around 1940. Most of this was done in secret at Park Orchards, near Ringwood on Melbourne's Eastern fringe, and was jokingly referred to by colleagues as 'Slutzkin's Death Ray'! His communications device, in its final incarnation, consisted of a very small torch bulb in a 'Lucas' signalling lamp housing, powered by an audio amplifier with treble boost to compensate for the thermal inertia of the filament. Reception was via 10 inch diameter parabolic mirror and gas photocell with an Ag-O-Cs (S1) photocathode.

Meanwhile, the German Military in association with the Zeiss optical concern developed several versions of their 'Lichtsprecher' device. This covert speech communicator modulated light beams by varying the total internal reflection angle of a prism with a secondary reflector spaced a half of a light wavelength behind it. This secondary reflecting surface was driven by amplified speech vibrations. As the modulated mirror varied in distance from the prism surface, the optical system would reflect varying amounts of light. A YouTube page showing the use of the Lichtsprecher device may be found here.

A more wholly electronic approach was taken by the United States Navy, which used modulated Cesium vapour lamps to communicate at infra-red wavelengths (see diagram below):

A resurgence of interest in optical communication came with the rapid advances in lasers and optoelectronics after 1960. In 1962, television signals were transmitted 18 miles using a modulated infra-red beam generated by a GaAs diode, prior to the general availability of the laser. The all-time distance record for terrestrial optical communication with speech modulation was set on 3-4 May 1963, when a 632.8 nanometer helium-neon laser beam was transmitted 118 miles by W6POP and W6QYY, from a point in the San Gabriel Mountains near Pasadena to Panamint ridge near Death Valley, California. An amplitude modulated 10-metre amateur radio transmitter was used for energising the laser. A web page detailing this 1963 expedition may be found here.

Since that time, research has centred around the pulse modulation of lasers (1963), coding techniques, heterodyne detection schemes using local laser oscillators (1965), optical FM (1968), and optical fibre light guide technology.

Optical communication is becoming successful as an engineering alternative to microwave technology owing to the development of the laser, the existence of an established optical technology, and the lack of success with millimetre wave hardware.

Atmospheric optical communication is likely to remain limited to non-commercial applications. These include amateur radio, citizen's-band type point-to-point communication, and perhaps local area community broadcasting, as proposed by the British 'Radio Love' group, and demonstrated between 1968 and 1971.

The commercial applications of communication at optical frequencies will almost certainly be in conjunction with optical fibre light-guides.

A major difference between radio and optical communication are the emergence, at optical frequencies, of quantum effects. For a given transmitter power, the number of photons generated will decrease as the frequency increases. This is predicted by the Einstein-Planck relation:

                                   Where E is the energy of one photon
                                              f is the frequency of the photon
                                         and h is a proportionality constant, called Planck's constant.

So many photons are generated for each watt of input power at radio frequencies that their propagation can generally be predicted by employing a 'wave model' to a first approximation. At optical frequencies, the photon effects can no longer be ignored. One has to stop thinking of the 'carrier' signal as being a wave, and start thinking of it as a stream of particles or quanta, whose arrival time at the detector is governed by probability theory. At these frequencies, the 'carrier' has two different frequency parameters:

To recover a useful signal, a communication system must receive at least 2.B photons per second, where B is the information bandwidth. This is for an ideal case where a detector will demodulate every photon, or in other words, where the detector will have 100% quantum efficiency. In practice, the number of photons per second required to extract a useable signal will be much larger, owing to noise sources and limiting background radiation. At optical frequencies, information bandwidth will usually be more limited by the received signal power than by the frequency of the carrier. Fortunately, the narrow beam divergence attainable with optical systems allow high signal intensities to be received at long distances.

Owing to the corpuscular nature of the received beam, random variations of the transmitted optical signal, with its statistical fluctuations of power, become a significant noise source at light frequencies.

Unlike propagation at radio frequencies where atmospheric gases are generally transparent, the atmosphere can seriously degrade optical signals through scattering, absorption, refraction and dispersion.

Scattering is caused by particles of smoke, fog or haze suspended in the atmosphere, and for optical communication it can be modelled in three ways:

(i) Rayleigh scattering, a transmission loss due to suspended molecular particles much smaller than the wavelength of propagation. This type of scattering is inversely proportional to the fourth power of the wavelength. Blue light therefore encounters about 16 times the amount of Rayleigh scattering that red light encounters.

(ii) Mie scattering, a transmission loss due to suspended particles comparable to or larger than the wavelength of propagation, such as those encountered in fog, smog and haze. Mie scattering is very complex to calculate mathematically, but is severest when the particle size is approximately equal to the wavelength of propagation. Hazy conditions are due to small dry particles in the atmosphere, and here the use of relatively long (IR) wavelengths can result in greatly reduced attenuation. Stable fogs, consisting of water that has condensed on salt nuclei are often encountered in coastal and maritime regions. Stable fog particles are large, and result in severe beam attenuation. Selective fogs (smog) in which water condenses around smoke particles are found in industrial areas, and the particles are quite small, allowing transmission at IR wavelengths.

(iii) Scattering of radiation from unwanted sources (eg. the sun) into the beam path, producing limiting background light levels. Two of the processes have been outlined above, and a third is the result of reflection from the buildings and foliage in the vicinity of the transmitter.

For almost all wavelengths less than 1.25 microns, including the visible spectrum, and distances in excess of about 2.5 km, scattering is the major contributor to path loss and background light level limitations.

Absorption is caused by the atmosphere's molecular constituents. Peaks in the atmospheric absorption vs. wavelength curve correspond to spectral lines of the atmosphere's component gases, and these bands may be as narrow as 0.1 nanometer. Care must be exercised in selecting the wavelength of an optical communication system suited to atmospheric propagation. Absorption characteristics may vary by as much as 20:1 for different wavelengths.

[Note by author, May 2005: the graph above was produced with the HITRAN (high resolution atmospheric transmission) software developed in the 1980s at the US Air Force Geophysical Laboratory. A 0.83 µm GaAlAs IR LED would emit over approximately a .02 µm wide band, spanning many of the absorption lines between 0.81 and 0.85 µm. Judging from this graph, atmospheric transmission would tend to be better at the long wavelength end of the GaAlAs LED output band, around 0.845 µm. A laser source for atmospheric communication, having a much narrower bandwidth, would need to have its wavelength chosen with much greater care!]

Fortunately, the visible spectrum is almost free of molecular absorption bands, as the atmosphere's major constituents, N₂ and O₂ absorb mainly ultraviolet radiation. Ozone (O₃) blocks UV radiation below 300 nanometres, while absorptions in the visible spectrum include minor O₃ bands between 500 and 700 nanometres, and oxygen bands at 688 and 760 nanometres. The most important absorbing compounds at visual frequencies and low altitudes are H₂O and CO₂. At infra-red wavelengths, owing to the high absorptions of O₂, CO₂ and particularly H₂O, the atmosphere is transmissive only in a series of narrow "windows", lying between the absorption bands and resonance frequencies of these compounds.

Refraction and dispersion: Atmospheric refraction fluctuations may bend, concentrate or disperse a transmitted light beam. Gases in the atmosphere are a non-homogenous mixture of 'cells' of varying density and temperature, moving with the prevailing winds. When the atmospheric density discontinuities or "cells" are large in relation to the diameter of the transmitted beam, a transmitted light beam may steer clear of the receiver entirely. A broad, dispersive transmitted beam from optics of large aperture will help to mitigate these problems. More often, atmospheric refraction only causes fluctuations in the received beam intensity. When the cells are small compared to the beam diameter, alternate dispersal and focussing of the beam may result, similar to the effect of 'twinkling' on stars. This may cause the transmitted beam to fade rapidly at rates up to 500 Hz. The effect is worst in hot, windy conditions at low altitude; and is the main reason for favouring pulsed-FM transmission technique [or digital communication] over simpler analogue intensity modulation for atmospheric optical communication. By using FM, amplitude variations due to atmospheric degradation may be clipped at the receiver.

In laser systems, atmospheric turbulence additionally causes a loss of beam coherence over distances of only a few tens of metres, with phase cancellation effects producing an extra source of noise and frequently rendering heterodyne reception by a local laser oscillator impractical. With coherent light, phase cancellations can cause the signal not merely to fade, but to drop to zero. Lasers have proven to be far more problematical for atmospheric optical communications than many of their theoretical advantages (narrow bandwidth and negligible divergence) might imply. The use of light guides and optical fibres seems to be the only way of overcoming these difficulties.

In spite of these apparent limitations, the author has found the reliability of atmospheric optical links to be surprisingly good, particularly below a 5 km. range. In our experimental system using high pressure mercury arc lamps and photomultiplier receivers between January and April 1976, a 5 km link gave 3 months of constant service when we used it on every other night or so, the signal to noise ratio never falling below 10 dB, even during heavy rain. Usually, the signal to noise ratio of this simple AM link exceeded 40dB.

At any time about 50% of the earth's surface is under cloud cover. The angular distribution of light scattered from clouds is a function of water droplet size and the wavelength of propagation. Owing to the spherical shape of most suspended water particles, the scattered light intensity will be greatest close to the direction of the transmitted beam, or normal to it (ie. back in the direction of the transmitter, a form of optical back-scatter).

Assuming that the beam widths are very much smaller than the angle between the transmitter and receiver beams and the line joining the two sites; and considering the simple case where the transmitted and received beams are tangential to the earth's surface with a cloud at the beam intersection, the minimum height of the cloud for small θ (in radians) will be:

If below H_min, the cloud will be below the horizon at both transmitter and receiver. If above, there will be a decreased scattered intensity.

Some special optical considerations come into play where 'cloud-bounce' or NLOS (non-line-of-sight) propagation is the mode employed. Beam dispersal is not such a critical consideration as it is with direct LOS operation, so that relatively small aperture optics can be used for transmitting a light beam up to the cloud. On the receive side, however, the photons reach the detector via dispersive specular reflection, so that the maximum area of receiver aperture is particularly critical. A military searchlight's large parabolic mirror would be ideal.

The reflection from clouds appears to be greatest at the high-frequency end of the visible spectrum - violet, blue or green light - so that the usage of modulated mercury arc sources, modulated carbon arcs, blue/green high-power Luxeons or even HeCd (violet light) lasers becomes desirable for transmitting. At these wavelengths, also, the photomultiplier retains considerable advantages as a detector, in photon efficiency, speed and large sensitive area. With a large photocathode area, an iris diaphragm in the focal plane, in front of the photomultiplier, could be used to instantaneously vary receiver beam dispersal to adapt to the changes in the clouds propagating the reflected beam. At a given receiving site, the transmitted beam will sometimes illuminate a small patch on a dense cloud, for which a narrow receiving beam would minimise unwanted pickup from extraneous sources. At other times, when the clouds are tenuous and the beam reflects from various depths within a cloud, the receiver beam should ideally be broader to accommodate the whole visible 'shaft' of reflected light.

At the receiver, polarising filters may provide some immunity from scattered skylight pollution, depending on the state of the atmosphere and the incident angle of sky dispersal from unwanted sources.

[Added note, May 2005: A unique problem associated with 'cloud-bounce' is the potential loss of high modulation frequencies caused by the 'blurring' of waveforms, owing to their multipath propagation from different thicknesses of cloud cover. The problem is discussed in an interesting paper by S Arnon and N S Kopeika of the Department of Electrical And Computer Engineering, Ben Gurion University of the Negev in Beer Sheva, Israel: 'Adaptive Optical Transmitter and Receiver for Space Communication Through Thin Clouds'. The paper was published in 'Applied Optics', Vol. 36, No. 9, pps. 1987 - 1993, 20 March 1997. This paper is believed to be available in .pdf form in the Internet.]

Most forms of cloud cover have some degree of translucency. Reflections of the transmitted beam from a range of cloud thicknesses and levels can be expected. The denser lower-level clouds, cumulus or nimbus (900 to 4000 metres altitude) could be expected to provide the least multipath hf loss, but unfortunately their low altitude would also provide the least potential increase of range over the horizon. Cirrus or alto-cumulus clouds, up around 12,000 metres are more tenuous, and frequently composed of ice crystals. The received signal would be of greatly reduced strength, and could be badly effected by multipath. [Note by the author, May 2005: The usage of the digital sub-noise level narrow-band transmission techniques such as 'JT44', 'JT65', MCW or similar would seem ideally suited to this work. Yves F1AVY, who currently holds what I believe to be the world's record for terrestrial non-line-of-sight optical communications, has used the JASON (12PHD) V099 program and the WOLF (KK7KA) program to communicate over 40 km by bouncing light off a hillside near Lyon, France].

Experimenters should be aware that the indiscriminate 'shooting' of high-intensity light beams up into the night sky is an aviation hazard and can be a punishable offence. The laws governing this are similar to the laws governing the usage of 'LIDAR', the optical range-finding system for determining cloud heights and upper atmosphere wind speeds. For specific experiments of this type, and especially for laser experiments, clearance from the local Civil Aviation authority should be sought. Vigilant 'spotters' armed with binoculars are usually required, by law, to warn of aircraft straying near the transmitted beam, and to shut it off if necessary.

These types of NLOS propagation can also apply to reflection from earth-based objects, such as trees, tall buildings, communications towers and so forth to reach a receiver site obstructed from the transmitter. One possibility is to reflect modulated light beams from local telecommunication towers or microwave dish radomes - though one suspects that by-laws may come into play here!

By far the greatest source of unwanted background ambient light in optical communications is the sun, whose radiation approximates that of a 6000⁰K blackbody. This daylight energy is received via reflection from the background surrounding the transmitting end of the link, and by scattering in the intervening atmosphere. Four methods for the reduction of this background ambience are:

(i) Reduction in the receiver's dispersal angle and its field of view. A compromise must be struck between narrow beamwidth and the ease of lining up. Receiver mounting stability can be a major constructional problem with the very narrow beamwidths achievable in optical systems, particularly with lasers. Even seismic disturbances can present alignment problems with the miniscule divergence angles possible via the usage of gas lasers. There is a definite practical limit to the reduction of transmitted beamwidths, set by the turbulence of atmospheric paths. The optical methods for controlling the transmitted dispersal angle of a light beam will be discussed in a subsequent section.

(ii) Reduction in the receiver's optical bandwidth via narrow-band spectral filters passing only the wavelength of propagation. For noncoherent sources generating radiation over a moderately wide range of wavelengths, a wide spectral filter such as a dye-based or photographic Wratten filter may be used to pass an appreciable amount of the transmitted light. Light emitting diodes, for instance, have a typical spectral line width of 30 nanometres, so that red LED's can have their received flux contrast improved against background by using a 'red 25' filter. This filter is used in black and white landscape photography to increase cloud contrast in a blue sky. One can also obtain short-wavelength cutoff filters from Edmund Scientific (address below).

Gas discharge light sources may be used with a narrower bandwidth interference filter to accept one of the dominant spectral lines emitted. For example, using a high pressure mercury arc lamp as a source, any one of the following wavelengths could be selected, according to the spectral response of the photodetector used:

The various wavelengths emitted by a typical high pressure mercury arc are shown in the spectrum above. Either of the dominant spectral lines at 435.8 or 546.1 nm could be selectively filtered to reduce background ambient pickup. A typical high-pressure mercury arc's relative power outputs of these wavelengths are tabulated below:

Wavelength (nanometres).	Approx. energy distribution.	Luminous flux (percent energy multiplied by eye response).
404.7 & 407.8 nm (violet)	13.3 %	0.1 %
435.8 nm (blue-violet)	20.4 %	0.7 %
491.6 nm (blue-green)	0.1 %	0.1 %
546.1 nm (green)	28.5 %	51.9 %
577.0 & 579.1 nm (yellow)	24.4 %	40.1 %
Continuous	13.1 %	7.1 %

Xenon or carbon arc lamps have a relatively continuous emission spectrum, and may not be selectively filtered in this way. This augurs against their usefulness for optical communications, in spite of their high intensity.

The standard type of narrow-band optical filter consists of a transparent film of known thickness with semi-reflecting metallic films on both sides of the pellicle (see diagram below). Maximum transmission occurs by a process of optical interference at the wavelength for which the dielectric's optical thickness is an integer multiple of half-wavelengths. Single or multilayer filters of this type are obtainable, covering any wavelength required between 200 nm (UV) and 20000 nm in the far-infrared. These are sometimes offered by scientific supply houses as 'laser line filters'.

For the visible spectrum, interference filter transmissions of around 70% are attainable for 20 nm bandwidth, 50 % for 10 nm and 30% for 1 nm. Most filters have a positive shift of maximum transmission wavelength with temperature, in better designs only about 0.02nm/C^o. Prices rise steeply with increased filter diameters. Typical 1994 prices from Edmund Scientific were US$39 for 12.5 mm diameter, US$75 for 25 mm, US$179 for 50 mm. For that reason, they're only economically viable for mounting directly in front of the detector or at the focal plane aperture of an optical receiver.

The main problem arising from the design of these filters is that they only provide their specified pass-band with incident radiation perpendicular to the filter surface. For radiation arriving at an angle to the filter, the pass-band shifts to shorter wavelengths, and broadens. At extreme angles the passband can become asymmetrical and misshapen, or it can split into two or more discrete bands. This is a particular problem with lenses of large aperture and short focal length, where rays can enter the filter over a broad cone angle. For this reason, they suit receivers using glass telescope objectives of with a numerical f/D ratio of f8 or higher, and are not really suited to very short focal length fresnel objectives.

For objective lenses of moderate focal length, a negative (concave) corrector lens can be used to make the rays parallel in the manner shown below - my thanks to Yves F1AVY for forwarding me this diagram:

For night operation, particularly in rural areas, filtration losses probably outweigh any potential for discrimination from unwanted sources. Nocturnal operation is frequently optimised by removing all types of optical filtration and running the detector 'nude'.

Interference filters for a range of visual and IR wavelengths are available from:

Spectrolab, 12484 Gladstone, Sylmor, CALIFORNIA (USA) [caution: 1979 address!]	Newport Precision Products, [based in California] c/- Spectra-Physics Pty Ltd, 2-4 Jesmond Road, Croydon 3136, Victoria, AUSTRALIA.
Edmund Scientific Company, 101 E. Gloucester Pike, Barrington, NEW JERSEY (NJ 08007-1380, USA.)	Infrared Industries Inc., Thin Film Products Division, 62 Fourth Avenue, Waltham, MASSACHUSETTS (MASS 02154, USA.) [Beware: 1979 address - check it!]

Heterodyne reception can also be used to reduce received bandwidth, but at optical frequencies this involves costly specialised optical mixing hardware and intermediate frequencies in the tens of GHz.

(iii) The use of long (IR) wavelengths to some extent alleviates scattering problems as the wavelength becomes larger than the Rayleigh scattering particles, but the atmosphere only transmits IR in proscribed narrow bandwidth windows, and radiation on these wavelengths cannot be viewed by the human eye to permit easy focussing or alignment. In practice, the difference between red light and near-IR in atmospheric transmission is not very great. A comparative test with the same optical systems, using red (660 nm) and IR (840 nm) LEDs undertaken by VK3KAU and the author showed that the IR transmitted 2 to 4 dB better over a distance of 43 km. The difference is barely noticeable, particularly with fast fades of about 7 dB depth, so that the convenience of visible light generally outweighs the few advantages of IR operation.

(iv) The scattered light of the sky is partially polarised, so that polaroid filters may be experimentally positioned at the receiver to remove this component of the scattered light. The necessary angular alignment of the polaroid filter will vary through the day with the changing position of the sun in the sky.

The lenses or mirrors used for transmission and reception in optical communications are analogous to the antennae used in radio communication. Ideally, the transmitted light beam should fall completely within the aperture of the receiver. This can't be achieved economically except over very short ranges, neither is it necessary, but the effects of the inverse square law can be offset quite effectively by optical means.

Firstly, consider the case of a laser beam which can be focussed to a 'source' point, the size of which is diffraction-limited to dimensions determined by the wavelength of the light:

The accompanying graph shows the loss of laser light between two collimating optical systems (transmitter and receiver) of equal diameter and aperture "a". The loss is seen to be kept low out to a distance "R" between the systems, of the order of a / θ, where θ is the divergence angle of the transmitted beam. For θ = 10^-4 radian (a dispersal so small as to render aiming difficult) and a = l metre, R approx = 10 km. Since the beam focussing achievable depends on wavelength, atmospheric turbulence and the aperture area of the system's optics; as θ_approx = wavelength / a , another way of expressing the distance for low loss with diffraction-limited optics of high quality is:

The usage of lenses of large aperture area and sources of high intensity are therefore shown to be the factors of paramount importance in long-range atmospheric laser communications.

The case of a collimated non-coherent light source, such as a gas discharge lamp or surface-emitting LED collimated by an imperfect glass lens or fresnel is more complex. The source is no longer diffraction limited, but occupies the physical dimensions of the source surface. Beam dispersal is governed by the angle θ formed by lines connecting the edges of the source with the centre of the collimating lens:

As shown above, the transmitting case enables us to geometrically determine the beam divergence, the beam intensity and the proportion of the transmitted beam intercepted by the receiving aperture (area A_R). These parameters are governed by the aperture size of the source (area A_S), the radiance of the source (L), the aperture of the transmitting lens (area A_T ), transmitting lens focal length (u), the distance to the receiver (l), the size of the virtual image of the light source at the receiver (area A_im) and the aperture of the receiving lens (A_R ). Without going through the steps of deriving the formula, demonstrated by Gagliardi and Karp in their book on 'Optical Communication Systems', the received optical power (Φ_R) is given by the following:

A source of high radiance (L) and high intensity is obviously required, as are large lens apertures at both the transmitter and the receiver.

In practice, the image formed by a collimating lens is always imperfect, and for non-coherent sources of appreciable source size, like LEDs or gas discharge sources, the collimator does not need to possess the accuracy of a telescope mirror or an achromatic telescope objective. It can be as simple as a mass-produced plastic fresnel lens, which can be obtained with enormous apertures for low cost. An excellent paper on the theory, manufacture and materials of plastic fresnel lenses is available at:

Nearly all of these fresnels are designed to have their grooved side facing the longer conjugate (ie the distant station) with the flat size inside the optical enclosure facing the neaby source or photodiode. [Author's note, May 2005: A 40 cm (16 inch) square fresnel, for instance, with a focal length approximately the same as its diameter, is available from the 3DLens Company in Taiwan for US$28.50. Refer their web catalogue ]:

Where - G is a geometric correction factor, determined by ^f/_D ratio (see graph below).

The geometric correction factor arises because as the focal length is reduced, the beam divergence increases at a greater rate than the total beam power: This aspect of optical system design is frequently overlooked in the standard texts:

Another aspect of optical design frequently overlooked in the standard texts relates to maximising the performance of an inaccurate collimation lens. With plastic fresnel lenses, the divergence of the transmitted or received beam often relates more to the accuracy of the lens surface than to its focal length. A plastic, non-rigid lens produced by a moulding process will invariably have surface irregularities causing the image of a point source at infinity to focus into a fuzzy patch, instead of the usual sharp point surrounded by Airy disc diffraction rings. The size of this focal patch will depend on the accuracy of the fresnel lens, and the degree to which it is corrected for spherical aberration. Many early fresnels, such as those in old overhead projectors, had no spherical aberration correction.

For moulded fresnel lens collimators, all other factors (eg source intensity) being equal, beam intensity will be maximised when the light source or detector aperture fills the fuzzy focal patch of the fresnel. The effective (virtual image) size of the source or detector can be varied to suit any given fresnel collimator by the use of a small secondary plano-convex lens of high diopter value (short focal length), as shown below. With an accurate glass collimating lens this arrangement would only increase beam divergence. With an inaccurate fresnel, the beam intensity will increase rapidly as the source aperture is effectively increased to the size of the fresnel 'fuzzy focus' patch. However, if the virtual source size is increased further, the beam divergence will increase and the beam intensity will slowly fall. The filling of the 'fuzzy focus' patch can be controlled by changing the relative distances of 'A' and 'B' in the optical system shown diagrammatically below, while maintaining a focussed image at infinity. In practice, observation of the transmitted beam intensity on a distant wall or screen can be used to optimise the lens spacings in the optical system below:

In the early 1980s, a team at the Bell Labs in America developed a laser receiver configuration using an innovative optical enclosure design. The same enclosure could be used as a transmitter with a Luxeon modulated light source. This design minimised the chance of pickup away from the main transmitting beam by placing a metal honeycomb or 'septa' structure in front of the lens. The inside surfaces of the septa and the interior of the optical enclosure behind the lens had to be painted matt black to absorb stray light. The honeycomb structure in front of the lens reduced the chance of concentrated sunlight being focussed inside the enclosure - a major safety concern with optical receivers. The enclosure's mass and wind-loading footprint was minimised by mechanically coupling the fresnel collimator to the detector with a light-tight truncated cone, in the manner shown below:

Duplex operation - the simultaneous operation of transmitter and receiver - is a major attraction of optical communications. Radio systems can rarely provide a transmitter and receiver working simultaneously on the same frequency. Optical systems offer this unique feature as a result of their high directivity, with the ability to convey conversations in both directions at once, or with digital links providing continuous error correction signals in both directions. Duplex operation demands separate fresnel lenses for the photodiode and modulated light source. Consequently, many optical transceivers resemble a huge pair of binoculars surmounted by some form of aiming device, such as a rifle sight. Since 1989, our own optical units have generally consisted of twin adjacent fresnel lenses clamped to a single protective glass cover, with a light-proof baffle separating the optical chambers behind them. With appropriate black matting behind the fresnels, aided by the liberal application of black silicone sealant, light leakage between the fresnels through the cover glass can be made insignificant. A typical transceiver of this type can be seen in the photos of Mike Groth VK7MJ, above.

The mounting of both lenses in a single unit simplifies optical alignment. By focussing the unit's light source onto a distant wall or chimney, one can align the rifle sight so that its cross-hairs are central on the distant illuminated spot. To account for parallax, the cross-hairs should be slightly offset on the distant spot's image by the same distance and direction separating the sight and the transmitting lens axis. Then, by modulating the transmitted beam with an audio tone, the receiver beam can be co-aligned by shifting the photodiode in the focal plane of the receiving lens. When the tone level received by reflection is maximised, the unit is optically aligned and the photodiode can be fixed in place. If the transmitter, receiver and sight are all in the same unit, alignment isn't easily lost. With the single optical unit, the work of lining up on the distant station is also halved.

However, in links operating in excess of about 30 kilometres, very large fresnel lenses are desirable. To keep the optical unit to a size transportable in an average car, and to keep wind loading down, single lens operation is best. By sacrificing the duplex option, one lens can still be used as a transceiver, but the modulated source and the photodetector have to alternate at the fresnel's focal point. They can be shifted back and forth into this position on a hinge or slide. This could be operated by a cable release, a solenoid or a pneumatic arrangement. Alternatively, the light source and photodetector can remain fixed while a hinged mirror behind the fresnel lens shifts the position of the focal point. The modulated light source would probably have to be switched off during reception periods in this type of unit, with the mechanical movement linked to the switching.

Our design - untested so far - is shown in the diagram below. It particularly suits large fresnels of moderately long focal length. By placing a mirror between the fresnel and its focal point, one can fold the optical path to occupy the minimum physical space. The mirror would be no more than a quarter of the area of the fresnel lens. The mirror could be hinged along one edge, tipping up or down to move the focal point onto the source or the detector, as required. If the mirror is tilted downwards, the modlight source and detector could be placed near the bottom of the fresnel inside the front of the unit, pointing backwards at the mirror. This positioning of the electronics would keep the device's centre of gravity low and stable, and the electronics would be enclosed and protected from the weather. Heat liberated by the transmitter heat sink could be directed to rise inside the unit, warming the fresnel slightly to prevent the formation of dew.

(i) Must have maximum aperture area to intercept the greatest possible number of incident photons from the distant transmitter.

(ii) Must have high directivity (i.e. resolution) to discriminate against light from extraneous sources. This is achieved by using a lens or mirror of moderately long focal length with a focal plane aperture as small as possible consistent with avoiding diffraction effects, to eliminate all parts of the image except that of the source aperture at the transmitting end of the system.

The receiver's focal plane aperture should also allow for the acceptance of light from the transmitter around the principal image resulting from lens inaccuracies or aberrations. A large aperture lens is essential for good resolution. Chromatic aberration is not a concern with most light beam links, as they are only required to operate over a narrow band of optical frequencies. Long focal length lenses have the advantage of being less subject to the effects of coma and spherical aberration, but a short focal length will keep the size of the optical unit to practical dimensions, suited to mounting on a photographic tripod. Compromises are necessary.

Some local optical firms sell 5" double-convex magnifying glasses, mass produced by moulding in Japan for about $5.00 each [note: 1970s prices!]. While their optical quality is not wonderful, they're adequate for amateur optical communication. Lens coatings to reduce internal reflections are not generally needed in this application. Coatings improve the total flux available by around 10% to 15%, amounting to less than 1 dB of signal. Focal length, and the ^F/_D ratio of the lens is, within extreme limits, not an important consideration in an optical receiver. For a given lens area, the same number of photons will be collected from a distant transmitter, irrespective of the focal length.

For the transmitting end of the link, we want the maximum possible beam intensity. However, directivity is not a critical consideration, and the beam may be allowed to disperse a little. This allows for beam bending by atmospheric refraction fluctuations, the limitations of optical unit mounting stability, and the effect of wind loading on the targeting stability of the optical unit. A fairly low ^f/_D ratio is desirable, though an ^f/_D ratio of less than unity will increase beam dispersal excessively without increasing beam intensity (see 'geometric correction factor' graph, above).

In our original 1976 optical link, the terminals were arranged as shown below, a moulded glass parabola being used for transmission and a 13 cm magnifying lens used for receiving:

Large diameter optics are desirable, not only for the maximising of optical gain, but also for maximising the cross sectional area of the coupling cone (or transmitted beam) between the transmitting and receiving apertures. This allows the optical system to average the disruptions caused by raindrops or atmospheric density cells traversing the beam. It also provides a degree of spatial diversity when birds, insects or other obstructions may fly through it. A narrow beam, such as that coming directly from a laser cavity, could be completely disrupted by a very small obstacle or atmospheric turbulence cell.

In selecting the light detector for a communication system, the frequency of operation is the critical determining factor. The use of infra-red light, with its fog penetrating properties and large number of available photons per watt would seem a desirable expedient. However, all detectors with infra-red sensitivity are also sensitive to thermal noise, ideally requiring cooling to realise maximum sensitivity. The difficulties of focussing and aligning an infra-red beam which the eye can't see also augurs to the usage of visible light, between 400 and 700 nanometres.

In spite of advances in semiconductor light detection technology, the photomultiplier is still an attractive device for the detection of weak visible light signals at room temperature. It is particularly useful at the violet end of the spectrum, between 350 and 500 nm.

While the silicon photodiode and the cadmium sulphide photoconductive cell both have higher quantum efficiency (the number of photoelectrons emitted from the photocathode per incident photon) than the photomultiplier in the visible spectrum, their internal noise and dark current at room temperatures is not easily overcome. These limitations of the available detection devices, together with the difficulties encountered in detector refrigeration (eg. window frosting, condensation, and potential cracking of the glass envelope) all augur towards the use of optical communication systems in the 400 to 500 nm region, at least for the amateur. [Author's note, May 2005: this statement was based on the feeble LED performance available in the 1970s. I would now use the new Luxeon III's (super high output visual LEDs) operating around 630 nm at the red end of the spectrum, capable of 4 watts output. P-I-N photodiodes are quite sensitive for receiving red light, but they lack the speed combined with the large sensitive area that photomultipliers had]. Suitable modulated light sources in the short-wavelength region of the spectrum include the mercury arc lamp, the HeCd laser and the argon laser (488 nm).

Photomultipliers require a 1000 volt power supply, but their associated circuitry is very simple, their internal gain very high (typically 5 x 10⁶ ), and their output is large. Response speeds reach about 50 MHz on standard designs, and may extend into the GHz region with special design. Photomultipliers are available with a range of different photosensitive surface materials, with different spectral response curves designated with 'S' numbers by the Electronics Industries Association (EIA). Some have their sensitive surface on an opaque metal plate, some on a semitransparent window on the glass end of the tube. The composition and characteristics of the various common photocathode types for photomultipliers are given in the following table:

	Cathode components	Entrance window material	Photo-cathode support	Peak quantum efficiency (%)	Luminous sensitivity (µA/lumen)	Peak response wavelength (nm)	Dark current @ 25^oC: amps/cm²
S-1	Ag-O-Cs	Visible light transm. glass	Window or opaque	0.43	25	800	10^-11 to 10^-13
S-3	Ag-O-Rb	Visible light transm. glass	Opaque	0.53	6.5	420	10^-12
S-4	Cs₃Sb	Visible light transm. glass	Opaque	12.4	40	400	10^-14
S-5	Cs₃Sb	UV transmitting glass	Opaque	18.2	40	340	10^-14
S-8	Cs₃Bi	Visible light transm. glass	Opaque	0.78	3	365	10^-14 to 10^-15
S-9	Cs₃Sb	Visible light transm. glass	Window	5.3	30	480	10^-14
S-10	Ag-Bi-O-Cs	Visible light transm. glass	Window	5.5	40	450	10^-13 to 10^-14
S-11	Cs₃Sb	Visible light transm. glass	Window	15.7	60	440	10^-14 to 10^-15
S-13	Cs₃Sb	Fused silica	Window	13.5	60	440	10^-14 to 10^-15
S-17	Cs₃Sb	Visible light transm. glass	Opaque reflecting	21	125	490	10^-14 to 10^-15
S-19	Cs₃Sb	Fused silica	Opaque	24.4	40	330	10^-14
S-20	Sb-K-Na-Cs	Visible light transm. glass	Window	18.8	150	420	10^-15 to 10^-16
S-21	Cs3Sb	UV transmitting glass	Window	6.6	30	440	10^-14
S-23	Rb-Te	Fused silica	Window	2	-	240	10^-17
S-24	Na₂KSb	Visible light transm. glass	Window	21.8	32	380	2 x 10^-17
S-25	(Cs)Na₂KSb	Lime glass	Window	12.7	160	420	^?
GaAs	GaAs	Visible light transm. glass	Opaque	10	250	830	10^-16

Several of the 'S' numbers missing from the table above relate to photoconductors (or light dependent resistors - LDRs - S16 = CdSe) and some to semiconductors (germanium = S14). In general these have a much higher peak quantum efficiency than any photoemitter (Ge = typically 43%; Si = typically 83%) but their lack of internal amplification (with the exception of avalanche photodiodes) and extended infra-red sensitivity generally makes them much noisier at room temperature. The semiconductors also require carefully designed external low-noise current amplifiers.

Typical multiplier photocathode surfaces suitable for use in the 400 to 500 nm region may have peak quantum efficiencies of around 16%, as this graph of popular photomultiplier spectral response curves reveals.

Like most photosensitive devices, some photomultiplier cooling is advantageous, though not obligatory. Variation of photomultiplier dark current with temperature for various photocathode substances is shown in the graph below. Significant reductions in internal noise may be made by cooling to the temperature of dry ice. Below -40⁰C little improvement can be attained with most photocathode materials.

Commercially available gas-discharge lamps used for industrial illumination are commonly based on:
                                                     (1) Neon gas.
                                                     (2) Xenon gas.
                                                     (3) Sodium vapour.
                                                     (4) Mercury vapour.
                                                     (5) Fluorescent (frequency conversion) systems.

Except for short-range work, fluorescent tubes and neon lamps are unsuitable, as these sources of large area and low intensity cannot be optically collimated into an intense beam. The neon lamp's total light output in commercially available versions is very limited; and the fluorescent lamp's response speed is limited by the persistence of glow in the phosphor coating.

The atomic processes involved in gaseous discharge reveal the relative merits of sodium and mercury lamps.

Light particles or 'photons' are absorbed by an atom when its outer electrons move to an orbit further from the atomic nucleus. The electron can move from its unexcited position, known as its ground state; or if previously excited, it may move to a larger 'permitted' shell. The amount that the electron moves is dependent on:

(1) The amount of energy absorbed. Photons of high energy will cause a large movement. As photon energy is proportional to frequency, blue light will cause a larger electron displacement than red light.

(2) As the electron must move to certain 'permitted' shells within a specified atom, only photons of energy equal to the permitted energy level jumps will be absorbed. Light can only be absorbed by the atom at those specific frequencies. This produces an absorption spectrum which is unique for every substance.

Conversely, if an electron loses energy by falling to a lower energy level, closer to the atomic nucleus, this energy loss can be emitted in the form of a photon. As with absorption, the light emitted by each substance will occur at a series of frequencies equivalent to the permitted energy level jumps for that substance - the emission spectrum of the substance.

An electron's transition from a given energy level to the ground state produces the 'resonance line' emission of the particular substance involved. At this resonance frequency, the gas is capable of selectively re-absorbing its own light output, converting it into transitions between higher energy levels, generally at lower frequencies. Selective absorption increases with the pressure of the gas in the discharge, so that to promote the emission of the resonance frequencies, the gas must be kept at low pressure. The resonance line is also suppressed at higher discharge current densities, as the atoms may then encounter successive excitation of incident electrons to higher energy levels before falling back to the ground state . Alternatively, the atom may transfer its energy to an electron without emitting a photon at all.

With a sodium vapour discharge, where the resonance lines fall within the visible spectrum at 589.0 and 589.6 nm (yellow light), the most efficient light outputs are achieved at low gas pressures and low current densities. For this reason, to provide a reasonably efficient light output, sodium vapour lamps are of low intensity and large source area, and therefore are unsuited to optical communications.

The super-high pressure mercury lamp offers many advantages for optical communication. Mercury's resonance lines are at 185.0 and 253.7 nm in the ultraviolet. To promote efficient radiation at visible wavelengths, the higher energy level transitions at 404.7, 435.8, 546.1 and 578.0 nm are promoted by the use of a discharge at high pressure and high current density. This promotes an intense, small source area well suited to collimation for optical communications.

In commercially available high pressure mercury lamps, the discharge is maintained within a small quartz phial. This is usually surrounded by a much larger diffusing bulb, sometimes coated with a fluorescent substance to make use of the residual ultraviolet output. For optical communication usage, the quartz phial can be removed from the diffuser by smashing the outer diffusing bulb, and re-mounting it inside a small clear-walled glass vessel or jar. This should be done with great care, and on no account should the arc be operated without its protective glass enclosure. The quartz phial should not be handled without gloves, and if it should happen to receive a fingerprint, the quartz should be carefully washed in denatured alcohol before usage. A physical hazard exists from the high pressures at which the arc operates, and it emits harmful ultra-violet radiation which penetrates the quartz bulb, most of which is blocked by glass.

The mercury discharge strikes at about 180 volts with a light blue glow initially filling the entire quartz bulb. As the bulb warms and the mercury sublimes to vapour at high pressure, the discharge narrows to a thin blue-green arc of high intensity. About 15 minutes are required for the lamp to reach its final intensity. In that condition, the arc can be modulated to about 50 % at 10 kHz.

A disadvantage of this type of light source is that after it has warmed up, its striking voltage increases so much that it is impossible to re-strike if it happens to extinguish on a modulation peak, unless it's left to cool for a few minutes. The use of a feedback circuit and a negative peak clipper in the modulator is suggested, to ensure that the arc is never completely extinguished. Nonlinearity of modulation may be corrected by positioning a photodiode near the arc, and connecting it to the modulator in a negative feedback loop.

The lamp must be de-rated to run in A.M. service with DC bias. To run at a continuous 30-watt output, for instance, at least a 60 watt lamp rating is required, to accommodate peak power output under fully modulated conditions. The de-rating must also account for cathode heating via positive mercury ion bombardment. With AC operation, where the cathode is switching rapidly from one end of the tube to the other, the heating effects are effectively shared between the two electrodes. With DC operation, most of the heating takes place at the cathode, increasing the dissipation requirements over that for AC operation. Arc polarity must be reversed from time to time to prevent the excessive ion bombardment of one electrode. To overcome this problem, RF bias could be tried, but we have found that the occasional switching of DC polarity via a DPDT switch is also permissible.

The modulator should include some method of controlling the DC bias current through the arc, and metering to measure arc voltage and current, as these parameters drift with changes of ambient temperature and modulation conditions. If arc current isn't monitored, it can drift upwards beyond the dissipation ratings of the lamp and the modulator tubes.

[Author's note 1 June 2005: Some years after the conclusion of our mercury arc experiments in April 1976, Rodney Reynolds VK3AAR suggested a potentially superior replacement to the vacuum tube modulator reproduced above. We had tried using power transistors in series with the arc, and invariably failed. The negative resistance of the arc was always the problem. As the collector current, together with the arc current rose, the resistance of the arc fell. The arc's voltage drop also fell. This would drag the collector voltage to a peak as the collector current simultaneously peaked. No transistor commonly available in 1976 was equal to the task of withstanding a peak of 5 Amps at 200 volts!

Rodney VK3AAR suggested the use of a modulator consisting of a high-voltage power transistor (eg. the BUX80) wired in parallel with the arc to circumvent these problems, with a hefty resistor dropping from the 200 volt rail to both, as shown in the diagram below:

The circuit above bears more than passing resemblance to the 'Heising control' or 'audio choke control' class-A modulators used for early broadcast transmitters in the 1920s. Please note that the above circuit has never been tested, so that prototype voltages and circuit values would probably have to be 'tweaked' in practice. The BUX80 would be dissipating a considerable amount of heat, so it would have to be bolted to a heatsink of generous proportions, and probably fan-cooled.

A possible improvement to the modulation linearity would be arranged by placing a resistor of about 1 ohm between the Hg arc and earth. The voltage developed across this resistor could be used as a source of negative feedback for application to a previous amplifier stage, to linearise the current drive through the arc. Extreme care should be taken with this circuit, as the voltages of operation are potentially lethal. Fast-blow fuses should be placed in series with the 220 volt DC rail.]

For amateur use, optical communication offers a cheap and license-free alternative to microwave systems for point-to-point communication. It can also be used for omni-directional transmission over short distances.

Simpler transmission systems, requiring less than, say, 50 KHz bandwidth, may use any readily modulated light source. Short range systems may employ light emitting diode sources which, except for the green or white phosphor-activated types, have a linear modulation characteristic, and are readily internally modulated at low voltages. Long range systems could use modulated high intensity gas discharge lamps, carbon arcs, gas lasers or solid state lasers. A coherent (ie. laser) light source is not mandatory, and may prove to be economically unjustified where bandwidth and limiting daylight is not a critical consideration.

Optical communication ranges through the atmosphere can extend to over 100 miles, and may be stretched beyond the horizon by the use of cloud scatter in favourable conditions. There has been little quantitative experimentation over these distances, in spite of the relative ease with which they can be achieved.

The system outlined here, using modulation of the power supply to a mercury arc, and a 6097 photomultiplier receiver, is only one of many alternative schemes for use at optical frequencies. Its effectiveness, in spite of its simplicity, indicates that the time is ripe for a substantial upsurge in amateur interest in such systems.

[Author's note 5 June 2005: A more modern optical transmitting system, using the new 'Luxeon III' LEDs capable of 140 Lumens' output will soon be posted on another web page on this site.]

Update: A number of web pages describing high-power LED-based optical transmitting/receiving systems may be found elsewhere at this, the modulated.org web site!

[Author's note, May 2005: This was written in the context of the old Wireless Telegraphy Act of 1903 (with amendments), which has since been repealed in favour of the Radio Communications Act of 1984]:

As this article was being written, I approached the Regulatory and Licensing Section of the Postal and Telecommunications Department and made enquiries regarding the licensing of the system outlined here. Mr. Ditchburn of the Victorian branch assured me that while there is no license covering such equipment, permits are available for such devices under the terms of the Wireless Telegraphy Act at no charge to the applicant. I have been given the verbal assurance that while my written application is being processed, I may proceed with my present experiments without fear of legal action.

[Author's note, 1 May 2005]: During the 1976 tests, various modulated sources and detectors were tried. LED's were tried and quickly rejected. Their output at that time was utterly insignificant by comparison with the arc. The available output flux from LEDs has increased by about five orders of magnitude since.

After my article's publication in January 1979, one boffin wrote to "Amateur Radio" virtually claiming that our system couldn't provide results comparable to the solid state systems of that time. He had used LED's and relatively small 35mm camera lenses, obtaining a range no greater than fifty feet. As I had not then received the optical transmitting permit from Telecom, I was unwilling to fully reveal the potentialities of the arc in terms of sig/noise and range, now revealed on another page on this website. The solid state system could then only beat the arc in bandwidth. I contacted the engineer concerned and challenged him to measure the sig/noise of our arc system in operation. The challenge was not accepted, neither did that engineer substantiate his own claims.

The situation only began to change in dollars for decibels in the late 1980s. Truly high output visual LED's dropped to mass-production prices at that time, as the prices of gas discharge tubes simultaneously rose. Laser diodes remained high in price until the mid-1990s, the cheapest that we could locate on the market in 1988 being a little over the Aust. $100 mark. Infra red LED's had been available with greater output power than visual LED's all through this time, but the wavelength of most of these coincided with the 930 nM H₂0 absorption band. Their transmission through the atmosphere is significantly poorer than visual LED's in most cases. The 830 to 850 nM GaAlAs IR LED's have much better atmospheric transmission characteristics. However, the practical difficulties of aligning and focussing an invisible infra-red beam gives visual LED's on 600 to 700 nM huge practical advantages in ease of focussing and alignment for amateur communication usage.

My interest in optical communication revived in 1987 as a result of the "Amateur Radio" magazine articles on the subject published by Mike Groth, then VK4CDG. Mike and I exchanged ideas, devices and optical components afterwards. The only aspect of Mike's thesis that I questioned were his range estimates for various optical systems. My experience in 1976 suggested that these estimates were excessively conservative. Tests with 660 nM, 2000 mCd red LED's and silicon PIN photodiodes confirmed my suspicions in April 1991, when we easily bridged 46 km across Melbourne between Hawthorn and Sunbury. Our usage of Luxeon sources since 2002, with resultant beam intensities increased by almost two orders of magnitude, provides further confirmation of the conservatism of Mike's range estimates. Worked in conjunction with large aperture lenses, Luxeons can be used to set up links literally as far as the eye can see.

A complete redesign in circuitry, optics and terminal unit mountings was demanded by the solid state components. The blue-sensitive photomultiplier was no longer a viable option for reception. PIN diodes are sturdier, smaller and almost as fast. They require a very low noise preamp in transimpedance (current-to-voltage converter) configuration, to provide a low noise, linear output. Their spectral matching to LED's is excellent, with a quantum efficiency exceeding 50% in their red-IR wavelength range. But the silicon photodiode noise is still higher than most photomultipliers, so the 'optical antenna' (collimating lens) used with them must have high gain.

Magnifying glasses of five inch diameter were used for the photomultiplier receivers in 1976. The cost of these glass lenses have risen beyond $50 in the past fifteen years. A much more viable option lies in the usage of the plastic fresnel lens. These are sheets of plastic with concentric grooves moulded on them, like an l.p. record. Each groove is a tiny refracting facet comprising an annulus section of a convex lens surface. Thus a large diameter, lightweight and inexpensive lens with remarkably short focal length is achieved. Fresnels are commonly used as condensing lenses in overhead projectors, and as "magnifying sheet" reading aids. In Australia, 10 by 8 inch magnifying sheets are available for about Aus.$12, and from a Taiwanese source (3DLens Company), a 16 inch square lens is available for US$28.50. The latter has over 16 times the collection area of our old 5" glass lenses, yet can be bought for half the price!

- Collaboration with field tests and arc modulator design December 1975 - April 1976 by courtesy of John Eggington, VK3ZGJ [now - May 2005 - VK3EGG].

- Graphs of photomultiplier response curves and noise dependence on temperature courtesy EMI Photomultiplier Applications Manual (1970) and Hammamatsu HTV Photosensitive Devices Catalogue, 1970.

- Bob Slutzkin VK3SK, "My 'Death Ray' Experiments", in OTN, the journal of the Radio Amateurs Old Timers Club of Australia, Melbourne, March 2000, pps. 12 - 14.

- A E Karbowiak, "Communication Systems in the Visible and Infrared Spectra, Present and Future" in Proceedings of the Symposium on Lasers and their Applications, IEE London, Sept-Oct 1964, pps 30-31. One of the earliest Australian papers written on the subject of laser communication systems, by a lecturer from the School of Electrical Engineering at the University of New South Wales, Sydney.

- A E Karbowiak, "Lasers and Optical Communication Systems", in 'Electromagnetic Wave Theory Part 1' edited by J Brown, pps. 419 - 439. (Proceedings of a Symposium held at Delft, The Netherlands, September 1965), pub: Pergamon Press, Oxford, UK, 1967. Professor Karbowiak's activity in the field of optical communications ceased when SDL in the UK decided to pursue optical fibres, rather than atmospheric communication, as a technology.

- Monte Ross, "Laser Receivers", pub: John Wiley & Sons, New York, 405 pps, 1966. The first comprehensive textbook on optical communication receivers for coherent sources of light. This is a solid text, meticulously referenced, now more valuable for its theoretical coverage than for its review of systems current in 1966, now chiefly of historical interest. The coverage of optical systems for unguided atmospheric communications is rather scant, otherwise this is a book of unprecedented usefulness. (Refer the picture of the author's well-thumbed copy, on the left).

- RCA Photomultiplier Manual, pub: RCA Electronic Components, Harrison, New Jersey, 1970 edition.

- RCA Electro-Optics Handbook, pub: RCA Solid State Division, Electro-Optics and Devices, Lancaster, Pennsylvania, 255 pps, August 1974 (reprinted May 1978).

- F E Goodwin: "A Review of Operational Laser Communication Systems" in Proceedings of the IEEE, vol 58, pps 1746 - 1752, October 1970.

- Fang-Shang Chen: "Modulators For Optical Communication" in Proceedings of the IEEE, vol.58, pps 1440 -1457, October 1970.

- Melchior, Fisher and Arams: "Photodetectors For Optical Communication Systems" in Proceedings of The IEEE, vol. 58 , pp. 1466 - 1486, October 1970.

- K Burlinson: "Modulated Light Communication" in Australian Electronics Exchange Bulletin (EEB), published by Reynolds and Gunther, Hobart, Tasmania, August 1968, February 1970, August-October 1972, and December 1972. Interesting experiments with mercury arcs and pulse width modulators by two young amateur radio operators active in Adelaide between 1967 and 1970.

- A H Sommer: "Photoemissive Materials: Preparation, Properties and Uses", John Wiley & Sons, New York, (256 pps.), 1968. This is the authoritative textbook on photoemitters, photomultipliers etc., written by one of the most prominent innovators in the field. Sommer was trained in Physical Chemistry in prewar Berlin, worked with the pioneering Baird Television concern in London in the 1930s, and ended his working life in America with RCA Laboratories. He was inventor of the S-10 and S-20 photocathodes.

Last update to this page: Friday 17 June, 2005 with minor additions made Tuesday, 18 January, 2011.