PHOTOPHONES REVISITED.

A review of amateur optical communications, originally published in 'Amateur Radio' magazine, pub: Wireless Institute of Australia, Melbourne, April 1987 pps 12 - 17; and May 1987 pps 13 - 17.

by Mike Groth, VK5AMG [now (2005) VK7MJ].

[Note by Chris Long, 7 April 2005: Just after Mike Groth published this article, he worked in nuclear medicine at the Brisbane General Hospital, and part of his duties involved the maintenance of whole-body scanners using low light level detectors, scintillation counters and the like. I wrote to him at that time, and the subsequent exchange of information began a loose experimental collaboration on optical communication that continued through three decades. 

Having used photomultipliers and modulated high pressure mercury arcs in the 1970s, I was initially sceptical that semiconductor technology could span longer distances through the atmosphere. Mike gave me a push into semiconductor sources and detectors and I was amazed at their improvement since my earlier optical comms experiments with VK3ZGJ (now VK3EGG) in 1976. Also in the late 1980s, really high output 'Stanley' 2000 mCd and 4000 mCd LEDs came onto the market. In 1991, I used Mike's semiconductor receiver circuits, fresnel lens collimators and the new ultra-bright 5 mm LED's to span 43 km from my old QTH in East Hawthorn to Peter Wolfenden VK3KAU's QTH in Sunbury. A decade later, with the advent of 1 watt 'Luxeon' LEDs in 2001, the modulatable light flux available increased by almost two orders of magnitude. With these 'Luxeons', Mike and I finally spanned 167 km in Tasmania on 19 February 2005.

Mike's original 1987 article gave very conservative estimates of potential operating ranges for optical communication systems. New light sources and large-aperture fresnel optics have improved on these distance estimates by almost a factor of ten.

We're putting Mike's original 1987 article on the net as a matter of technical and historical interest, as the principles and calculation methods are still perfectly applicable to current optical communication systems. In some ways, this publication marked the commencement of serious Australian amateur work in this field. Its comprehensive nature also makes the article a model of its type:]  

INTRODUCTION

Apart from limited military applications, optical telephony remained a relatively impractical form of communication from the invention of the photophone in 1880, to the development of semiconductor light sources and detectors in the 1960s. While optical fibres have become a major component of modern telecommunications, and infra-red remote controls are incorporated in many domestic appliances, optical communication has been largely ignored by radio amateurs.

Construction projects for photophones have been published from time to time over the last 60 years, but there have been few reviews of optical communication and its potential as a medium for amateur voice and data communication. This article is a mixture of history, theory and personal experience, written with the intention of introducing optical communication to the general body of radio amateurs and possibly stimulating further experimentation in the oldest branch of wireless.

HISTORICAL DEVELOPMENT

Early Developments, 1878 - 1918

The invention of the selenium cell in 1872 and the telephone in 1876, made it possible to detect modulated light, and Mr A.C. Brown of London is generally credited with the first transmission of articulate speech over a light beam in 1878. Much of the pioneer work in optical telephony was carried out by Alexander Graham Bell and Charles Sumner Tainter during 1879 and 1880, which was presented in a paper1, read by Bell to the American Association for the Advancement of Science in August 1880.

The Bell photophone (Figure 1) used a flexible plane mirror mounted at the end of a speaking tube, so that the sound pressure caused the mirror to change shape, modulating the beam intensity of the reflected light. The receiver was a selenium cell mounted at the focus of a parabolic reflector, and coupled to a battery and telephone receiver. Using this apparatus, Bell transmitted speech over a distance of 213 metres using sunlight, and shorter ranges were covered using various lamps as a light source.

Interest in photophones appears to have been dormant until the turn of the century, when German and Austrian experiments with current modulated carbon arc lamps, led to the production of a military photophone by the Siemens-Halske company in 1917. This unit used a current modulated carbon arc transmitter, and a selenium cell receiver, to give a night range of about 8 km. The German Navy was reported to have used voice modulated searchlights for ship to ship communication up to a distance of 7 miles (11 km).

The British were also active in photophone research during the First World War, and the vibrating mirror modulator was developed by Rankine as part of a research project for The Admiralty in 19162. Other methods of producing modulated light, including current modulation of carbon arcs and fine filament lamps were found to have very poor modulation characteristics. 

The selenium cell was the only photoelectric detector available until the development of the thalofide (oxidized thallous sulphide) and molybdenite detectors in 1917. These had a lower noise level than selenium and a faster response to infra-red radiation. 

An experimental photophone was developed in the U.S.A. by the Case Research Laboratories in 1918, which used a pressure modulated acetylene lamp (Figure 4) in the transmitter, and a thalofide cell with a valve amplifier in the receiver. A clear night range of 8 km was claimed with 24" (600 mm) reflectors at each end.

1919 - 1935

Improvements were made to optical modulators and detectors in the 1920s, by motion picture engineers developing the optical sound tracks on movie films. Photophones became a technical novelty for display at industrial exhibitions and science fairs, with the occasional construction project in the popular radio magazines.

Military Photophones 1939- 1950

There was a renewed military interest in optical telephony in the 1930s, and the German Army introduced the Zeiss Lichtsprecher infra-red photophones in 1935. The light source was a tungsten filament lamp with an infra-red transmitting filter, which was modulated by a vibrating mirror (or prism in the Li80). The receiver used a lead sulphide detector with an infra-red filter, and a valve amplifier. They were virtually unaffected by daylight, with a clear weather range of 3 km for the Li 50/60, to nearly 14 km for the Li 250/130.

The Japanese Army visible light photophone incorporated a vibrating mirror modulator and a caesium photocell detector, with an operating range of about 1 km in daylight and 2.5 km at night. An Italian Army photophone used a current modulated filament lamp as the light source, but few details appear to have been published outside of the military reports.

Both the German and American Navies used high pressure vapour lamps as modulated infra-red sources for navigation, identification and short range communication. The Germans employed mercury arc lamps of 500 to 2000 W, while the Americans developed the caesium arc lamp. Some military laboratories continued the development of high pressure arc lamps for optical communication until the 1950s.

Post War Amateur Developments

From 1945, the occasional letter appeared in the amateur journals describing experiments with current modulated light globes, but with the development of transistors and photodiodes there was a small but scattered group of amateurs experimenting with photophones in the 1960s. Most equipment used current modulated torch globes and phototransistors to transmit distorted speech. but some optical links using gas discharge tubes could transmit high fidelity speech and music.

Following the invention of the laser and infra-red light emitting diodes. there was an increased amateur interest in optical transmission between about 1966 and 1972. when several speech and video contacts, were made over distances of 100 km or more. Despite the rapid advances in the commercial application of optical communication since 1970, there has been little serious interest in extending amateur radio into the optical part of the electromagnetic spectrum.

OPTICAL THEORY

It has been assumed that the readers of this article have a basic understanding of optics, including the properties of lenses and mirrors. A simple description of some more advanced optical concepts has been included to assist in the later discussion of light sources, detectors, and optical systems.

Light may be loosely defined as electromagnetic radiation having a wavelength between 300 nm (3 x 10-7m) and 3µm (3 x 10-6m), which corresponds to a frequency range of 1014 to 1015 Hz. This definition includes visible light with a wavelength between 400 nm and 700 nm, as well as the long wavelength ultra-violet and near infra-red parts of the optical spectrum as shown in Figure 2. Optical communication systems usually operate in the visible or near infra-red.

Light is emitted and absorbed in small discrete energy quanta called photons. The energy carried by each photon is determined by its frequency or wavelength according to the formula;

The spectrum of a light source reflects the energy of the excited electrons, and the thermal electrons in a hot body emit broad-band radiation, whose dominant wavelength is a function of the absolute temperature as shown in Figure 3. The 2500°K curve is representative of the spectrum of the white light from a filament lamp or incandescent gas mantle.

The monochromatic light from a sodium vapour lamp, neon globe or led, has most of its radiant energy concentrated into a limited range of wavelengths, determined by the differences in the atomic energy levels in the source. A monochromatic light source has some advantages in an optical communications system, as it allows the receiver to be tuned to the transmitter's wavelength.

The short wavelength limit for an optical link is set by atmospheric absorption of ultra-violet wavelengths below 300 nm, and the long wavelength limit is set at about 3 µm by thermal background radiation and rising detector noise. Glass lenses and windows are transparent to wavelengths from 350 nm to nearly 2.5 µm, while quartz will transmit infra-red to 3.5 µm. Most transparent plastics are suitable for infra-red operation out to a wavelength of 2 µm (2000 nm).

OPTICAL TRANSMITTERS

Optical Intensity.

An optical transmitter generates a beam of intensity modulated light, either by modulating the intensity of a light source, or by passing the light from an unmodulated source through an optical modulator. In either case, the effectiveness of the transmitter is a function of the transmitter's beam intensity, and the depth of modulation.

Because light sources have a finite size and do not radiate equally in all directions, four parameters (see Figure 4), are used to describe optical brightness and intensity. These are;

FLUX (F)                        The optical power (watts).

INTENSITY (I)                The power radiated per unit solid angle in a given direction. (watts.steradian-l).

ILLUMINATION (E)        The optical power per unit area. (watts.metre-2).

LUMINANCE (L)                The intensity per unit source area. (watts.metre-2.steradian-l).

For a point source of intensity I radiating equally in all directions, the total flux radiated is 4.π.I watts.

Visible light photometry is based on a white light standard (the candela), and visual brightness comparisons between light sources. The unit of luminous flux is the lumen, and a light source with a luminous intensity of 1 candela, is emitting one lumen of visible light per steradian. The candela replaces the older unit of the candle-power, originally based on the intensity of a sperm wax candle.

A watt of green light at the wavelength of peak response of the human eye (555 nm), is equivalent to a luminous flux of 692 lumens. The luminous efficiency for light of other wavelengths is reproduced in Figure 5, which may be used to estimate the radiant power from luminous flux measurements.

Transmitter Optics.

The simplest form of optical transmitter consists of a modulated light source mounted at the focus of a lens or mirror as illustrated in Figure 6. The intensity of the transmitter beam is given by;

                                                                             G .D2
                                               Ibeam =            __________ . Isource
                                                                               ds2

Where G is a geometric correction factor for the  f/D ratio of the optical system (Figure 7). Provided the focal length is not too short, the output lens (or mirror) will have the same luminance as the source, and the beam intensity will be a function of the source luminance and the lens area.

The divergence of the transmitter beam (θb) is determined by the ratio of the source diameter and the focal length. The use of a more intense source with the same luminance will increase both the power and divergence of the transmitter beam, but the beam intensity will remain unaltered. An optical system with a very low f/D ratio (such as a deep parabolic reflector) will give a very high beam power, but it can be seen from Figure 7 that the beam intensity will be less than that produced by a lens (or mirror) of the same diameter and moderate focal length. This apparent contradiction arises because the beam divergence increases at a greater rate than the total beam power as the focal length is reduced:

A very narrow beam can make the transmitter difficult to align, especially in an infra-red system where the beam is invisible. For an optical transceiver, the transmitter beamwidth should be wider than the receiver's field of view, so that the transmitter will be correctly aligned when the receiver is aimed for the maximum signal.

Modulated Filament Lamps.

A tungsten filament lamp has a high luminance in the visible and near infra-red (typically 105 W.m-2.sterad-l), but the poor modulation of the light output (Figure 8), reduces the effective modulated luminance to the order of 100 W.m-2.sterad-l

Despite the low depth of modulation and considerable distortion, current modulated torch globes were widely used in amateur photophones for voice communication over distances up to a kilometre on a clear night.

Gas Discharge Lamps.

Low pressure gas discharge lamps, including neon bulbs and fluorescent lamps, can be modulated to 10 KHz or more, but their luminance is very low (typically 10 to 20 W.m-2.sterad-l ). A gas discharge has a non-linear relationship between voltage, current and light output, but speech and music can be reproduced with reasonable fidelity using pulse.

High pressure sodium and mercury vapour lamps are widely used for floodlight, factories and street lighting, and are readily available with power ratings from 70 W to 2000 W. The luminance (typically 6000 W.m-2.sterad-l ) is almost independent of the wattage rating, and lamps of the 100 W size would be suitable for amateur experimentation. The audio modulation characteristics of these lamps is not known, but published data indicate that better than 50% modulation of the light output could be expected for frequencies up to 5 KHz.

The main disadvantages of high pressure lamps are the relatively high cost, limited life (500 - 2000 hours), and the long warm-up time. Sodium and mercury vapour lamps require at least 10 minutes operation to evaporate the metal in the lamp and produce their full light output. An optical transceiver with a high pressure vapour lamp, would have to run its transmitter continuously, with a shutter to cut the beam off during reception.

Light Emitting Diodes.

Light emitting diodes are junction diodes made from compounds of gallium, aluminium, arsenic and phosphorus. which emit nearly monochromatic light when forward biased. The emission wavelength depends on the chemical composition of the diode crystal. and ranges from 930 nm in the near infra-red for gallium arsenide (GaAs), to blue light at 500 nm for aluminium phosphide diodes. 

The light emitting diode is the most convenient light source currently available for amateur optical communication. The output is proportional to the forward current and may be modulated to frequencies exceeding 1 MHz. The optical properties of several common light emitting diodes are summarised in Table 1:

It can be seen that the efficiency and power output of a LED decreases with the emission wavelength, and an infra-red emitting diode has much greater output flux than a green LED for the same drive current. A high intensity red LED is a suitable modulated light source for demonstrations and experiments, as the visible radiation simplifies the optical adjustments.

High powered GaAs and GaAlAs infra-red emitting diodes are available with peak output powers of several watts, but the luminance of the source is probably not significantly higher than for smaller diodes. The efficiency and power output of an LED is temperature dependent (Figure 9), and some form of heat sinking is necessary if operating a diode near its maximum current.

Most light small emitting diodes are supplied in a transparent plastic package with a domed top, which acts as a lens and increases the intensity of the light along the diode axis. The lens does not increase the source luminance, but generates a bright halo as illustrated in Figure 10:

The effective luminance may be estimated by assuming the source diameter is equal to the diameter of the diode.

Mechanical Modulators.

A variety of mechanical devices have been devised over the past 108 years to impress voice modulation on a beam of light. As it impossible to cover these in detail, this review has been restricted to the basic principles of some of the more successful designs.

The intensity of a light beam may be modulated by altering the optical flux in the beam with a variable transmission device, or by changing the divergence of the beam. The latter approach was adopted by Bell in his 1880 photophone (Figure 1), which used a flexible mirror to vary the divergence of the reflected beam in sympathy with the sound pressure.

A modern version of the Bell modulator may be constructed by mounting a sheet of aluminised plastic or a thin glass mirror in front of a loudspeaker as shown in Figure 11. There should be a good seal between the loudspeaker rim and the mirror to achieve a tight acoustic coupling.

A simple modulator for use with a small filament lamp is drawn in Figure 12, where the flexible mirror and the lens form an optical system of variable focal length. The optical path from the lamp to the lens should be slightly shorter than the focal length, so that the filament will be in focus at the maximum concave curvature of the mirror. This modulator is most effective with a torch globe having a short narrow filament.

The flexible mirror is not a linear modulator, and the distortion rises rapidly with increasing modulation depth. Up to 30% modulation is possible with a very flexible mirror, but a transmitter using a glass mirror is unlikely to achieve more than about 5% modulation of the beam intensity.

The vibrating grid modulator is constructed from a pair of identical grids, each having equal transparent and opaque strips. One of the grids is fixed, and the other is attached to the voice coil of a loudspeaker driver as shown in Figure 13. The two grids have a static displacement of half a strip width, and driving the voice coil with an audio signal will modulate the transmitted light power about its quiescent value of one quarter of the incident optical flux.

The performance of the system will depend on the fineness and accuracy of the grids, as well as the mass and frequency response of the moving grid. The grids with strips about 1 mm wide could be a pair of photographic transparencies, or etched from a thin sheet of metal. The vibrating grid concept was independently suggested by Alexander Graham Bell in 1880, and by Sir William Bragg in 1915, but it was impractical with the acoustic drive systems available at the time.

The problems associated with the moving grids were overcome by Rankine in 1915, by using fixed grids and an optical lever as illustrated in Figure 14. The grids were located at the radius of curvature of the concave mirror, which formed an image of the first grid in the plane of the second. A small rotation of the mirror will move this image over the second grid, and modulate the luminance of the image formed by the second lens. The light from this image is collimated by the output lens to produce the main transmitter beam.

The rotation of the mirror may be produced by a high speed galvanometer, or a loudspeaker voice coil via a lever and fulcrum. Despite its greater complexity, the oscillating mirror modulator was the most successful mechanical design, and was used by the Japanese and Germans in their military photophones during the 1930's. Several other mechanical modulators have been developed using internally reflecting prisms or interferometers with movable plates. They have not been included in this review as they are precision devices which would not be suitable for amateur construction.

Electrical and Magnetic modulators.

The Kerr cell is a glass cell fitted with parallel electrodes and filled with nitrobenzine, which becomes doubly refracting in an electric field. The cell is mounted between a pair of crossed polarizers (Figure 15}, whose planes of polarization are at 45° to the electric axis of the Kerr cell. In the absence of an electric field no light is transmitted by the second polarizer. When a voltage is applied to the electrodes, the Kerr cell becomes doubly refracting, and the light emerging from the cell is elliptically polarized. As this now has a polarization component aligned with the second polarizer, some will be transmitted.

The optical path difference between the two polarization components in the cell is proportional to the square of the applied voltage, with a response time of less than 1 ns. Very strong electric fields are required to open the shutter, and a Kerr cell is often operated with an RF drive, when the light will be chopped at twice the excitation frequency.

Caution must be exercised when experimenting with Kerr cells, as very high voltages are involved, and nitrobenzine is very poisonous. It is also a powerful solvent and will attack most plastics, and a fatal dose can be absorbed through the skin.

A magneto-optic modulator (Figure 16), utilizes the Faraday rotation of a beam of polarized light shining along a magnetic field. Most transparent materials exhibit a very small Faraday rotation. and the effect is strongest in ferro-magnetic materials. An experimental voice modulator was developed in the 1960's using a thin section of yttrium-iron-garnet. which is transparent to near infra-red and exhibits a large faraday rotation.

Lasers.

A laser is a monochromatic light source in which the electron transitions have been synchronized by optical feedback, so that the photons are in phase with each other, and the light is coherent. Coherent light has the properties of a continuous wave, with a very narrow spectral bandwidth.

Lasers are best known for their high optical power output. Gas lasers producing over a kilowatt of optical flux are in regular use in industry for cutting cloth, wood and metals. The argon laser is widely employed for surgical operations, and solid state lasers with peak output powers of a terawatt (1012W) or more, probe the atmosphere and measure distances to satellites.

The most common laser for optical communications is the semi-conductor or diode laser, which is a modified infra-red emitting diode that generates coherent radiation. The luminance is much higher than a normal infra-red emitting diode, with a very narrow spectral spread. The infra-red is emitted with a divergence of about 10°, and can be current modulated to several MHz. [Chris Long notes, 8 April 2005: Laser diodes were very expensive when Mike wrote this article, but they are now commonly available from around US$15+, in a range of wavelengths from 635 nm (red) extending upwards into the infra-red. At longer wavelengths they can operate with considerable output powers, but in the visible spectrum where they're used as bar code readers and lecture pointers they mostly operate at about 1 to 15 mW. A few very expensive visible laser diodes attain 750 mW. They are not inherently collimated, radiating in an elliptical cone pattern, at an angle angle of about 30 degrees parallel to the junction, and around 8 degrees perpendicular to the junction. They are therefore usually supplied with a collimating lens somewhat less than 1 cm in diameter. For atmospheric optical communication this beam diameter would have to be spread optically in the manner of Figure 17 for eye safety and to reduce far-field divergence. Their beam cross-section is not very homogenous in brightness and they are only moderately coherent. Gas lasers still have the advantage of being inherently collimated and highly coherent in output (ie a high proportion of their radiation has a narrowly monochromatic output with the light waves in phase, plane parallel and minimally diverging - but they are much more expensive than diode lasers, they require an EHT supply and are difficult to modulate internally).]

The other common laser to which amateurs are likely to have reasonable access is the helium-neon gas laser, which emits up to 20 mW of red light with a wavelength of 632.8 nm. The light is emitted in a thin parallel beam, and the He-Ne laser is widely used in teaching, science, engineering, and surveying. The gas discharge may be powered by a DC current or an RF signal, and a 10 metre AM transmitter can. be used as a exciter for photophone experiments.

The parallel beam of light emitted by a laser will start to diverge after a short distance as a result of diffraction, but this can be reduced by expanding the beam through an astronomical telescope as depicted in Figure 17:

The diffraction spreading for a l00 mm diameter beam of coherent red light is about 15 microradians, but an expanded laser beam is observed to diverge at nearly 200 microradians (200 mm/km or 1 ft/mile), probably as a result of atmospheric turbulence and imperfections in the telescope.

OPTICAL RECEIVERS

An optical transmitter generates a beam of intensity modulated light. which is received by a photodetector and converted directly to an audio frequency electric current. This is similar to the early days of amateur radio. when incoherent signals from spark transmitters were received by crystal sets. Experimental coherent fibre-optic receivers have been demonstrated in several research laboratories. but a coherent optical communication system for atmospheric transmission is not likely to be available for some time.

Detector Theory.

A photodetector is a quantum device which uses the photon energy of the light to excite electrons and generate a current proportional to the energetic photon flux. All photon detectors have a cutoff wavelength λc which corresponds to the minimum photon energy required to excite an electron in the detector. In an ideal detector. each incident photon with a wavelength less than λc will liberate an electron, but the quantum efficiency of a real detector ranges from 0.03 to 0.5 electrons/photon.

The sensitivity of a photon detector is the detector current generated per watt of incident optical flux. It is inherently wavelength dependent (Figure 18), with the maximum sensitivity at a wavelength slightly shorter than λc. Radiation with a wavelength longer than λc will not be detected. The short wavelength limit is usually determined by absorption in the detector window.

A detector will generate white noise from electrical leakage, thermal excitation, and background light. The dark current is proportional to the square root of the detector area. and increases rapidly with the temperature and cutoff wavelength. The thermal noise contribution from a detector with a cutoff wavelength in the visible part of the optical spectrum will generally be less than the amplifier noise. Detectors sensitive to far infra-red radiation have a very high thermal noise level at room temperatures, and are not particularly suitable for optical communications.

Unmodulated light falling on a detector will generate white noise from statistical fluctuations in the photon flux. The light noise is proportional to the square root of the detector current, and is a function of the total light flux. Background light may be the main noise contribution in an atmospheric optical link operating during the day or on a moonlit night.

Receiver Optics.

In a typical amateur photophone receiver, the light from the transmitter is concentrated on the sensitive area of the detector by a lens as illustrated in Figure 20, although mirrors become more convenient if a large collector is required. The lens (or mirror} should have a focal length longer than its diameter for efficient light collection. Magnifying glasses or magnifying sheets make suitable receiving lenses up to a diameter of 250 mm, for visible or near infra-red signals.

The lens (or mirror) will form an image of the transmitter output aperture at the focal plane, which for a lens of reasonable focal length. will have a diameter of less than 1 mm. As this is smaller than the sensitive area of a practical detector, all the transmitter light falling on the receiving lens will fall within the active area of the detector. The detector current will therefore be proportional to the area of the lens or mirror, and independent of the focal length or the detector area.

A receiver will detect light arriving within a conical field of view, whose angular diameter is defined by the focal length and detector diameter. This field of view may include unmodulated light from scattered sunlight or moonlight, as well as modulated light from street lighting and other sources. The unmodulated light will generate white noise in the detector, while street lights and house lights will produce a strong 100 Hz interference.

As the noise and interference produced by the background light will increase with the receiver beamwidth, the receiver's field of view should be reasonably narrow. However, a very narrow field of view will make the receiver difficult to align, and may require some form of optical tracking system to compensate for changes in atmospheric refraction.

A detector about 2 mm in diameter will give a beamwidth between 3 and 10 milliradians (0.2°- 0.6°) with typical receiver lenses, which appears to be a reasonable compromise between interference suppression and ease of aiming. Larger detectors should have their effective diameter reduced with a focal plane aperture plate.

Photocells.

A photocell (Figure 21) is a vacuum diode whose cathode is coated with a material that emits electrons when exposed to light. The spectral response is determined by the cathode coating, which may be a mixture to produce a more constant sensitivity across the visible spectrum. Most photocathodes are relatively insensitive to red and infra-red light, but a photocell with a caesium cathode can detect infra-red wavelengths out to nearly 1300 nm.

Photocells are large detectors, with cathode areas from 1 cm2 to 10 cm2, but they have a very low thermal noise, wide dynamic range, and fast transient response. They have been successfully employed as detectors in visible light photophones in the past, but have been largely superseded by the silicon photodiode.

Photomultipliers.

A photomultiplier is a vacuum photocell fitted with a series of dynodes (Figure 22), which multiply the photocurrent by secondary electron emission. A typical photomultiplier has a sensitivity of the order of 105 amps/watt, and can detect a modulated light flux of 10-13W.

The photomultiplier is best suited for detecting faint light signals in a dark environment, and will saturate with a relatively low level of background light. They are very expensive ($50+) and relatively fragile devices, which can be damaged if exposed to a bright light with the HT applied. They are mainly used for amateur optical DX experiments, and are not recommended for inexperienced amateur experimenters.

Photodiodes.

A photodiode uses the photon energy to produce charge carriers in the depletion region of a semiconductor junction, and generate a current. This phenomenon is observed in several semiconductors, but the highest quantum efficiency and lowest leakage are obtained with a p-i-n junction, which has a wide depletion layer. A photodiode acts as a current generator, but is often operated with a reverse bias (Figure 23), to improve the transient response.

A silicon photodiode will detect ultra-violet, visible and near infra-red radiation out to a wavelength of 1100 nm. The peak response at 950 nm is near the emission wavelength for infra-red diodes, and many of the small photodiodes sold by electronic component suppliers have an integral infra-red filter. Photodiodes are well suited for optical communications, being small, cheap and rugged, with a high quantum efficiency, and a relatively low thermal noise level.

Measurements made by the author indicate that a BPW50 silicon photodiode connected to a low-noise audio amplifier can detect a tone modulated signal of 2 x 10-11 W at a wavelength of 900 nm. An AM speech signal of 10-10 W is quite readable, while an FM subcarrier system requires a signal flux approaching 2 x 10-10 W.

A germanium photodiode has a cutoff wavelength of 1800 nm, with its peak response at 1550 nm, and is a good spectral match for detecting the light from an incandescent lamp. The noise level is higher than a silicon photodiode, and an OAP12 germanium photodiode requires a flux of 3 x 10-10 W to produce a readable speech signal.

A light emitting diode may be used as both the light source and detector in a short range photophone, as shown in Figure 24. The cutoff wavelength of a LED operating as a photodiode is about the same as the emission wavelength, and the quantum efficiency is rather low when detecting radiation from another LED of the same type:

Photodiodes sensitive to far infra-red wavelengths have been developed using new semiconductor compounds with a very narrow energy band gap. These include indium arsenide (InAs), indium antimonide (InSb), platinum silicide (PtSi), and mercury cadmium telluride (HgCdTe), which is sensitive to radiation out to 15µm. Many of these detectors have been developed for military applications, and a lot of the technical data is classified.

Far infra-red wavelengths are of limited use for optical communications due to the high thermal background radiation between 3 µm and 50 µm. Detectors operating in this wavelength range have to be operated at about 80° K, which requires a liquid nitrogen cooling system.

Phototransistors.

Light falling of the base region of a transistor will generate charge carriers, which are multiplied by the transistor action. Silicon photodiodes with a cutoff wavelength of 1100 nm are readily available from electronics retailers, and are widely used in optical isolators and position sensors. Germanium phototransistors may be obtained by removing the opaque black paint from an older glass encapsulated germanium transistor such as an OC70, OC71, or OC75.

A phototransistor is often operated with an open circuit base for maximum sensitivity, but this produces a high noise level, as the leakage current and background light photocurrent are amplified together with the signal. The dynamic range is limited, and the transistor will saturate at moderate levels of background light. By operating a phototransistor in a bootstrapped amplifier as shown in Figure 25, the quiescent current is stabilized by the DC feedback, while the base impedance is very high at audio frequencies. This circuit is relatively insensitive to background light, but can detect a tone modulated optical flux of 200 pW (2 x 10-10W), and a speech signal of about 1 nW (10-9W).

Germanium phototransistors (eg OCP71) were widely used in amateur photophones in the 1960's to detect light from modulated filament lamps, but were largely rendered obsolete by the development of silicon transistors. A germanium phototransistor has a high leakage current and noise level, but when operated in a circuit similar to Figure 25, it should be possible to detect a speech signal of less than 10 nW.

Photoresistors.

Several semiconducting materials exhibit a reduction in bulk resistivity when exposed to light. Since this is mainly a surface effect, a typical photoresistor is manufactured from a thin layer of photoresistive material mounted on an insulating substrate between a pair of conducting fingers. The resistance changes are relatively slow, and there may be some treble cut when detecting a speech modulated signal.

Photoresistors are the oldest form of photoelectric detector, and were used in all photophones until the development of the photocell in the 1920's. The selenium cell was the primary detector until 1917, when it was superseded by other materials, including thallous oxy-sulphide (Thalofide), molybdenite, lead sulphide, and cadmium sulphide.

Photoresistors are the noisiest class of optical detectors, and inferior to photodiodes for visible or short wavelength infra-red. The lead sulphide (PbS) cell with a cutoff wavelength of 3.4 µm is useful for detecting radiation at the long wavelength end of the near infra-red. The noise level is very high at room temperatures, and it operates best at -30°C, when a speech signal of about 10 nW (10-8W), can be detected. Dry ice, which sublimes at a temperature of -49°C is a suitable cooling medium.

The cadmium sulphide photoresistors sold as light dependent resistors (LDRs), are sensitive to visible light, with a relatively slow transient response. While they can detect a speech modulated optical signal with reasonable fidelity, they are much noisier than photodiodes and most phototransistors, and are not particularly suitable as photophone detectors.

Photoresistors using doped germanium are used for detecting very long wavelength infra-red radiation. The cutoff wavelength depends on the doping element, and varies from 25 µm for copper, to nearly 100 µm for gold doped germanium. These detectors are usually operated at about 4°K with liquid helium cooling.

OPTICAL LINK PERFORMANCE

Atmospheric attenuation.

The optical power in a beam of light transmitted through the atmosphere will decrease exponentially with distance as a result of scattering and absorption. Atmospheric attenuation is often the dominant factor in determining the range and reliability of an atmospheric optical link over distances of a kilometre or more.

Provided the distance is large compared with the diameter of the transmitter lens (or mirror), the illumination (E) produced by the beam at a distance R is given by;

The atmospheric attenuation coefficient is the sum of three main components, namely rayleigh scattering from fine aerosols, absorption by atmospheric gases, and scattering from large suspended particles such as fog, dust and thick smoke.

Rayleigh scattering describes the scattering of energy by particles smaller than the wavelength, such as air molecules and fine aerosols. The scattering decreases with the fourth power of the wavelength, and is responsible for the blue colour of the sky and the blue haze observed over mountains. Red and infra-red light will penetrate haze better than blue light, but the transmission losses due to rayleigh scattering are relatively low.

Ultra-violet radiation is absorbed in air, and unsuitable for optical communications except over very short distances. Quartz windows and lenses are required, as glass is opaque to wavelengths shorter than 350 nm. An ultra-violet optical link would present a significant visual hazard to anyone looking down the transmitter beam without eye protection.

Atmospheric water vapour produces strong absorption bands in the near infra-red, as can be seen in Figure 26, which shows the transmission factor over a 1 km path on a fine autumn or spring day. The infra-red absorption would be lower on a clear frosty night, but more than doubled for a humid summer's day.

[Note by Chris Long, 8 April 2005: It would be easy to misinterpret the graph above, as the depth of its absorption dips are partly a function of the averaging produced by the graph's poor wavelength resolution. If one narrows the bandwidth of the measuring light source to obtain a higher resolution graph of atmospheric transmission, it would have sharper and more numerous dips in its transmission curve, like the graph below. Figure 26A is based on a  computer simulator program developed by the United States Air Force in the early 1980s, into which various figures of distance, humidity etc can be fed to provide a prediction of transmission at various wavelengths for a wide range of conditions. For the graph below, the CSIRO Atmospheric Physics Division in Aspendale (Vic.) was kind enough to feed various atmospheric 'scenarios' into their computer for us, to predict a fair range of possible results under differing weather conditions. The graph below would be typical on a clear, dry day, and has a wavelength resolution similar to that of LED sources:

[Even the graph above, sufficiently accurate for non-coherent sources of relatively narrow bandwidth like LED's (around 30 nm wide), is nowhere near the fine wavelength resolution required to predict the atmospheric transmission of something like a gas or ruby laser. The graph below was produced by 'tuning' a ruby laser, which can be done by varying the temperature of its emitting ruby rod - see the graduations in degrees Kelvin below. A similar high-resolution atmospheric transmission model for the whole of the visible and infra-red spectrum is now believed to be available in the USAF's "HITRAN" freeware computer program. The incredibly numerous and sharp atmospheric absorption wavelengths are a troublesome factor in predicting the 'transparency' of the atmosphere for a given laser's output. To produce a reproducible atmospheric transmission figure, frequency control via fine temperature control of the laser rod would be essential. These problems do not occur so much with non-coherent sources.]

[We now revert to Mike Groth's original article - C.L.]:

Atmospheric absorption from rain, mist, fog, smoke or dust, is the main limitation on the reliable operating range of an optical link, and there is no significant difference in the transmission of infra-red and visible light in fog or rain. It is not possible to predict the signal loss due to adverse weather conditions with any precision, but a rough estimate of the attenuation coefficient may be made from the daylight visual range with the aid of Table 2.

Background light.

Background light falling on the detector will generate white noise, which is often the main noise contribution in an optical receiver operating during the day. The detector current from the background light will be a function of the brightness of the background at the operating wavelength, the receiver beamwidth, and the spectral response of the detector and optical filters.

The reflectance and colour of the background will depend on the transmitter environment (trees, sky, buildings, etc.), and the ambient illumination will vary with the weather and the time of day. However it is possible to estimate background light levels for special cases, so that the daylight performance of different systems can be compared.

The albedo or average reflectance of the earth is about 0.3, and the solar illumination at the surface is 1100 W/m2. It has been assumed that the background at noon on a fine day has a brightness of 330 W/m2, which is equivalent to a luminance of 50 W m-2 ster-1. The corresponding spectral radiance R, is plotted in Figure 27:

[Note by Chris Long, 7 April 2005: I have added Figure 27A, reproduced immediately above, to give a more detailed view of expected daylight background radiation. During most of the daylight hours, solar radiation at sea level peaks around the wavelength of green light - 500 nm - which, perhaps not surprisingly, roughly matches the peak response wavelength of the human eye. The solar irradiance drops off rapidly at the red end of the spectrum and it falls even further in the near infra-red. Solar background radiation drops to 30% of the green wavelength peak at 1000 nm. The light scattered by the sky falls even more rapidly with increasing wavelength, and the sky appears almost black in infra-red photographs. The amount of undesired daylight background radiation detected by an optical receiver will vary according to the background 'seen' around the transmitter site by the receiver's optical system. Green foliage reflects infra-red light strongly while sky light does not. A simple red Wratten filter (sold by photographic dealers for black and white special effects as the 'red 25' or 'red 25A' filter) placed over the Si PIN diode will therefore help to reduce this daylight background, and a narrow-band filter centred on the transmission wavelength may be even more effective. However, communication systems should avoid the wavelengths shaded on the graph above, which represent atmospheric absorption bands mostly attributable to the spectral lines or resonance bands of H2O, CO2, O2 and 03 (ozone). These obviously reduce the solar irradiation received at ground level on those wavelengths, which might be desirable, but these bands would also absorb a modulated light beam travelling horizontally near sea level to a far greater degree. 

At night the light loss caused by an optical filter will often outweigh its contribution to signal recovery, and optical filtration then ceases to be useful. Back to Mike's narrative...]

The background luminance on a heavily overcast day, may be less than 10 W m-2 ster-1, and at sunrise and sunset, the solar illumination is about 1% of its noon value. The full moon is about a million times less bright than the sun, and the background radiance in these cases may be estimated by dividing the values read from Figure 27 by the appropriate factor.

It can be shown that the flux reaching the detector of an optical receiver operating with a background radiance R.λ, is given by;

It can be seen that an optical receiver for operation in the presence of background light should have a narrow field of view, and an optical filter centred about the operating wavelength. A very narrow band interference filter (less than 3 nm bandwidth) can be used with a gas or injection laser, but a wider bandwidth (around 30 nm bandwidth) is required to transmit the radiation from a light emitting diode. A simple red or infra-red filter will give a significant improvement in the signal to noise ratio when detecting radiation from an incandescent light source.

Typical values for the background light flux and noise level for several common detectors and filters, are given in Table 3. This table assumes a lens diameter of 100 mm and a focal length of 250 mm. For lenses or mirrors having a significantly different f/D ratio, the NEP from the table should be multiplied by 2.5 D/f.

The table above assumes:           1. 100 mm diameter lens or mirror.
                                                  2.  f = 250 mm
                                                  3. Full noon sun. Lbg = 50 W m-2 ster-l

For other weather or lighting conditions, the detector flux and noise power can be estimated as follows;

Overcast day:                 Divide Wbg by 10.                    N.E.P by 3.

Sunrise or sunset:            Divide Wbg by 100.                 N.E.P by 10.

Full moon at zenith:         Divide Wbg by 1,000,000.       N.E.P by 1000.

Moonrise or moonset:     Divide Wbg by 108.                 N.E.P by 10000.

Operating Range.

The operating range of an optical link is dependent on the weather and time of day, and any quoted range must be qualified with the appropriate operating conditions. The vacuum range is the theoretical communications range in the absence of atmospheric absorption, and is a convenient parameter for expressing the optical performance of a given transmitter and receiver. The operating ranges for various conditions can be estimated from the vacuum range and the atmospheric attenuation coefficient, as illustrated by the following example.

A simple photophone transmitter consists of a current modulated Tandy XC880 GaAlAs infra-red diode mounted at the focus of a 100 mm magnifying glass (f = 250 mm). The transmitter beam intensity may be calculated as follows.

The specifications for the Tandy XC880 IR LED are:
                                         Emission wavelength = 880 nm
                                         Power output @ 20 mA = 1 mW
                                         Dispersion angle = 24°, (half power point)
                                         Source diameter (ds ) = 5 mm

The source intensity is calculated by assuming the radiation is emitted into a 24° cone, when:

The transmitter intensity and beamwidth will be;

If the detector is a BPW50 infra-red photodiode. then a signal flux of 10-10 W will be required for speech reception. For a 100 mm diameter lens. the minimum transmitter illumination will be;

The operating range (OR) can be obtained from the equation:

                        Log10(VR)  =  Log10(OR) + OR x [Transmission loss (dB/km)] /20

This equation does not have a simple analytic solution for the operating range. but can be solved by successive approximations. From table 4. the clear air transmission loss at a wavelength of 880 nm is 0.8 db/km. which gives an operating range of 7.4 km on a clear night.

In the middle of a fine sunny day, the background light noise for a BPW50 infra-red photodiode at the focus of a 100 mm lens (f = 250 mm), would be about 2 x 10-10 W (Table 3). In this case, a signal flux of the order of 10-9 W will be required for speech reception. Repeating the previous calculations with Wmin = 10-9 W, will give a vacuum range of 4.3 km, and a clear air daylight range of 3.3 km.

The background light noise at sunrise and sunset will be about the same level as the detector and amplifier dark noise. Assuming a total noise level of 5 x 10-10 W, and a minimum useful signal of 2 x 10-10 W for speech communication, the clear weather twilight range would be 5.9 km. The background light from a full moon would produce a detector noise of less than 10-12 W, which is much smaller than the typical receiver dark noise, and moonlight will not significantly effect the operation of this optical link.

The effect of water vapour may be illustrated by repeating these calculations for an optical link using a CQY89 GaAs LED as the light source. The intensity of the CQY89 (7.5 mW/sterad @ 50 mA), is similar to the XC880, giving a vacuum range of 13.7 km. The emission wavelength of 930 nm, is on the edge of a water vapour absorption band, and the predicted operating range is reduced to between 4 km on a humid summer evening, and 6 km on a frosty night. The predicted clear daylight range is 2.2 km to 2.9 km, depending on the humidity.

The estimated clear weather operating ranges for various optical communications systems are listed in Table 5, and it can be seen that quite simple equipment can transmit speech or data over distances of 2 or 3 km in clear weather. Over long optical paths, the received signal strength is primarily determined by the atmospheric attenuation, and very intense transmitter beams are required for long distance optical communication.

  Reliability.

The reliability of an optical link depends on the path length, as well as the frequency and severity of adverse atmospheric conditions. Signal losses of 200 dB/km at a wavelength of 900 nm have been observed by the author during thick fog, with fluctuations of 30 dB/km over periods of a few seconds. Under these conditions, an optical link using infra-red LEDs and photodiodes would have an operating range of about 180 metres. A modulated gas laser beam would be readable at 500 metres, and a transmitter using a 100 W quartz-iodide lamp would have a range approaching 280 metres.

Therefore an optical link using simple components can provide reliable communication over distances of 100 to 150 metres in all weather conditions. Signal dropouts would be experienced during heavy fog at 200 metres, while a light fog or heavy rain would disrupt communications over a 1 km optical link. Depending on the equipment used, amateur photophone contacts of 5 to 10 km could be expected on clear nights, with possible DX contacts of 50 km or more under suitable atmospheric conditions.

OPTICS, RADIO AND WIRELESS.

The use of light to transmit information was a form of wireless communication under the broad definitions employed in the Wireless Telegraphy Act, but the 1983 Radio Communications Act defines a radio transmission as:-

(a) any transmission or emission of electromagnetic energy of frequencies less than 3 terahertz without continuous artificial guide; or

(b) any highly coherent transmission or emission of electromagnetic energy of frequencies not less than 3 terahertz and not exceeding 1000 terahertz, without continuous artificial guide.

This definition excludes incoherent optical signalling systems, such as amateur photophones or infra-red remote control systems, but a commercial laser powered optical link is a radio system, and requires a licence.

At present there are no Australian frequency allocations above 300 GHz, and it would assist in the orderly development of the sub-millimetre spectrum, if the W.I.A approached the Department of Communications with a proposal for amateur allocations above 300 MHz. This application could include reasonable use of coherent radiation from 100 THz to 1000 THz (3 µm to 300 nm), for amateur communications experiments.

CONCLUSIONS

Optical communication is a practical method for transmitting information over short distances, and is used commercially for computer links between city buildings, or across roads, where it is not practical or economic to use a wire or radio circuit. Optical links would be well suited for linking amateur computers, especially between apartment blocks where RF links can cause interference with adjacent entertainment and security systems. An optical packet message system would be tolerant of the occasional signal dropout caused by rain, fog, or birds flying through the beam, and it is not impossible to visualise the future establishment of an amateur optical packet network in the high-rise residential systems of the capital cities.

Optical DX can provide a challenge to the radio amateur or experimenter, who likes to do things the hard way. Optical voice and data transmissions of 100 km or more have been achieved in the past, and optical moonbounce is technically possible. In many ways, optical communication has come of age after a century of retarded development, and is both the oldest and one of the newest branches of amateur radio.

ADDITIONAL READING

This review is a distillation of information gathered from many sources by the author over a period of 18 years, and it would be impossible to give a comprehensive listing of references. Much of the theory can be found in standard physics textbooks, but the following references make interesting background reading, and provide a suitable starting point for a detailed literature search if desired.

1. A.G. Bell: "On the Production and Reproduction of Speech by Light" American Journal of Science, Vol 20, No 118, Pages 305- 324, (October 1880)

2. A.O. Rankine: "On the Transmission of Speech by Light" , Proceedings of the Physical Society of London, Vol 31, Pages 242- 268, (1919) 

3. W.S. Huxford and J.R. Platt: "Survey of Near Infra-Red Communication Systems" Journal of the Optical Society of America, Vol 38, No 3, Pages 253- 268, (March 1948)

4. H.S. Snyder and J.R. Platt: "Principles of Optical Communications Systems" Journal of the Optical Society of America, Vol 38, No 3, Pages 269- 278, (March 1948)

5. N.C. Beese: "Light Sources for Optical Communication", Infrared Physics, Vol 1, No 1, Pages 5- 16, (1961)

6. B.G. King et al: "An Experimental Study of Atmospheric Optical Transmission" Bell System Technical Journal, Vol 62, No 3, Part 1, Pages 602- 625, (March 1983)

------------------------------------

Back to Modulated Light DX

Last update to this page: Friday June 3, 2005

Page Count since 201005: