OPTICAL TELEVISION BROADCASTING

Project Notes



SKIP TO THE LATEST NEWS....


 

I welcome feedback! EMAIL me with your comments and if they are interesting I will include them here, giving due credit

As things (hopefully) progress, most of the detailed information will be published here. However some details, which could be politically or commercially sensitive, will not.

You can get up to speed by reading through the internet discussion board thread HERE. I now want to bring all the important information together in one place - on this page. I would like to thank the moderators of that board and all those who are contributing to the discussion.

Also I think I should include an apology. This page is undoubtedly something of a 'mish-mash'. It's not a polished account of the optical broadcasting project (still far from finished) - more an account of my thoughts as they come to me and a record of activities as they occur. As you'll see, I've been learning as I go along. I've found the process fascinating - I just hope it's not too annoying to read.

So what's this all about?

 

OPTICAL BROADCASTING is the broadcasting of programmes over free space using light rather than radio. In its current conception this is not point-to-point communication, but radiation from a mast in a flat omnidirectional beam. Invisible infra-red light would most likely be used.

One application will be to carry analogue Vintage Television on 405-lines. The signal modulated on the light will be a complete System A television channel - actually Channel 1 (vision 45 MHz, sound 41.5 MHz), not to be confused with baseband video and audio. As a result the supporting hardware has to be very fast. The big advantage of this is, at the receiving end, the demodulated signal is ready straight away for plugging into the aerial socket of a set. In this way, true television broadcasting has been achieved without any need for a radio spectrum licence.

Other services might also be carried alongside. For example, it should be possible to create a complete 'private' version of the radio spectrum, medium wave, short wave, you name it, all encapsulated on a beam of light. The current emphasis is on developing this for an amateur, analogue application. However there's no reason why commercial digital services might not be included too. The method lends itself to extremely large data downloads at locations where competing infrastructure is not available.

 

You can see a PowerPoint about this, picturesque but now very out of date... HERE.

Point-to-point transmission, using red light amplitude modulated with 45.0 & 41.5 MHz AM carriers to carry 405-line television and sound, was successfully demonstrated at the NBTV Convention at Loughborough University on April 9th 2016. There was a high speed red LED at the transmitting end and a very fast photodiode at the receiving end. The resulting picture of Test Card 'C' (with tone) was displayed on a 17-inch receiver. Quality was excellent.

The good news is, by using a low feed impedance, it has proved possible to modulate many other LEDs at 45 MHz. This removes a major obstacle. However, it has been found the modulation depth of 'ordinary' LEDs at VHF can be limited. The emphasis has now switched to using modulated lasers.

 

PROS

  • Requires no radio spectrum licence - regulation stops at 275GHz.

  • In principle, there would be room for a facsimile of the entire radio spectrum, operating simultaneously.

  • Channel B1 could be radiated, encapsulated in light.

  • Simple to receive - ready-to-go RF channel emerges from photo-sensor receptor.

  • Can provide a usable service area, given a suitable transmitter location.

  • A vintage set could be used as the receiver without modification.

  • Omnidirectional radiation could be limited above and below main beam, just like a radio transmission.

  • In principle, optical carrier has massive space for other one-way data applications, with commercial potential.

  • Near infra-red easier to generate and detect than far infra-red or millimetre waves.

  • ‘DC’ component of daylight could be prevented from flooding receptor sensor by filter. In any case, the existing receptor works in day or night.

  • Receptor units could use adapted satellite dish hardware, probably chromium plated.

  • Height of pick-up units would often not be critical.

 

CONS

  • Line of sight only - although reception would be possible over a wide area, it would be highly 'granular' in nature, especially in built up areas.

  • Badly affected by rain and fog. The use of infrared would slightly reduce this characteristic, particularly in mist. In my opinion this restricts the idea to intermittent data downloads and a 'special occasion' service for motivated viewers - such as vintage television enthusiasts!

  • Would not work when receptor was facing the rising or setting sun! In addition there could be a fire hazard if the sun were brought to a focus by the plated satellite dish receptor. This might mean the service coverage would have to be restricted to locations north and and south of the transmitter, with large sectors to west and east excluded.

  • 'Baseband Flutter' scintillation effect could occur on signal in fringe areas due to rising warm air and instability in atmosphere.

  • Quantum noise effect is greater on weak signals with higher modulation frequencies like these.

  • For infrared, thermal noise could be a problem.

  • Selective atmospheric absorption could badly affect a narrow-band laser carrier, if this were used.

  • Could be affected by interference from street lights, though their operating frequency and colour makes this unlikely...

 

UNKNOWNS & OTHER FEATURES

  • Provision of sufficiently fast LEDs or modulated laser array for radiating ring .

  • Legalities regarding aircraft lights on towers (to be gleaned at an appropriate time).

  • Hazard to vision of staring at emitter at close range (unlikely with envisaged radiation pattern except from nearby helicopter).

  • Bandwidth, sensitivity and noise floor limitations of available photo-sensors (on the way to being solved).

  • Heating aspect of near-infrared light at emitter's working power level (now not expected to be significant).

  • Need to ensure adequate vertical thickness of beam from radiating array to cover local topography.

  • Transparency of lens glass if infrared used (so far, seems not to be a problem).

  • Suggestion to use additional visible light LEDs at transmitter to assist receptor aiming during setup and to encourage 'helicopter pilot's' eye pupil contraction.

  • The emitter to be RF screened to prevent any standard radio emissions. The screening will likely take the form of an earthed mesh, and this will double as lightning protection.

  • The degree to which scintillation will be a problem with remote reception. A broad, or edge-concentrated radiating array might mitigate this.


THE 'NUMBERS' AS PREVIOUSLY ENVISAGED. CERTAIN INFORMATION IS NOT INCLUDED HERE. THE DATA BELOW RELATES TO AN INFRA-RED MULTIPLE LED BASED PROPOSAL, REPLACING THE SINGLE RED LED USED FOR A PROOF OF CONCEPT DEMONSTRTION ABOVE.

  • Theoretical number of LEDs: 8257.

  • Principal Wavelength: 850 nM.

  • (Measured) minimum acceptable signal pickup: 10 uW.

  • Theoretical optical transmitting power required: 578 watts.

  • Radiant flux per LED: 70mW.

  • Omnidirectional beam vertical spread: 6 degrees.

  • Vertical form factor of emerging beam: 0.105.

  • Radius of coverage: 5 km.

  • Depth of beam at fringe: 525 m.

  • Area of beam at fringe: 16.35 sq km.

  • Standard receiving dish diameter: 60 cm.

  • Area of dish: 282744 sq mm.

  • Gain of 60 cm dish (and focal collector lens) with S5973 photodiode: 61dB.

  • Radius of 1mW "safe" limit from transmitting array (assumes omnidirectional point with 6-deg vertical fan): 5.88 m.

  • Practical inner safety limit distance: 20 m?


IF YOU WANT TO DISCUSS THIS PROJECT WITH OTHERS, I RECOMMEND THIS THREAD ON THE NBTV BOARD.

OTHER COMMENTS (received by email) ALONGSIDE MY OWN NOTES AS THINGS PROGRESS, WILL APPEAR BELOW. SORRY! THESE NOTES ARE IN THE ORDER THEY CROPPED UP IN... ALL THE 'BLIND ALLEYS' ARE INCLUDED.

merlinmaxwell over on the UK VRR Board writes:

Fascinating, what is needed is an optical capacitor to stop the 'DC' ambient light.

Fortunately, not so. The sensor circuit uses a fast photodiode and opamp and seems very resistant to ambient light interference. I've had equally good results in daylight as in night-time conditions. However, if I had used a photomultiplier instead, for its sensitivity, 'flooding' of this type would have been a big problem... and destroyed the photomultiplier.


The next step is to build a higher power 'module' based on infrared LEDs and a bought-in 1 watt RF amplifier. This will be driven from the existing rig and will require a 75-50 ohm matching network and attenuator between the two.

Because the new infrared LEDs have their own built-in lenses, I am no longer planning to get a tower obstruction light to play with, using this as the transmitting array housing. Instead, I expect to eventually mount thousands of LEDs in a 'radiating ring'.

At some point I must investigate tower light legalities. I have a contact who runs a Business Radio station in London.

Later, to buy a satellite dish and try to adapt it as a receiving array.


CHRIS LONG from Australia is the long established guru in this field. His site is at www.modulatedlight.org. To date, his emphasis appears to have been more on narrow band (audio) point-to-point communication rather than wide band 'television broadcasting' as such. He has considerable achievements to his name.


There's a useful WIKIPEDIA page on this technology HERE.


Maybe I'm crazy for bothering to do this with AM modulated analogue television. There is no commercial potential unless a bulk data service is included. However a vintage television service would be 'fun'. The real geezer with a practical commercial data-carrying application is HERE.


THERE NOW FOLLOWS A SORT OF 'STREAM OF CONSCIOUSNESS' UNDATED 'BLOG', SETTING OUT EVENTS AS THEY UNFOLD. THE LATEST DEVELOPMENTS ARE AT THE BOTTOM.

You probably already have gathered that I favour the use of 'plated' satellite dishes as the 'receptors' in this project. These can offersubstantial gain along with being based on readily available hardware with proven resistance to windage.

I have been considering the limitations on the directions it will be permissible to point the dish, due to direct solar irradiation at one time of the day or other. Such irradiation could focus the Sun on the receptor box, causing it to smoke or catch fire! The following discusssion assumes an installation in the Northern Hemisphere.

Almost all installations will involve the dish pointing horizontally (unlike for satellite work) toward the remote terrestrial transmitting tower. So, our principal concern will be to avoid the rising or setting sun. This can be up to 23 degrees away - in either direction - from due east and west, according to the time of year. In addition, in temperate zones in winter the Sun gains and loses altitude in the sky only slowly, making the 'forbidden' zones to the south-east and south-west more extensive than those to the north-east and north-west.

The final factor to consider in calculating the guard band is the field of view of the plated satellite dish as seen from the finite size of the pickup receptor. For example, would the Sun, even if off-axis, still catch the pickup receptor and burn it?

The focal ratio of most satellite dishes lies between f0.28 and f0.42, making them very 'fast' by normal optical standards. Of course, the aberrations on the image they will produce will be far greater than those from a respectable lens or mirror, but that doesn't matter as much in this application. The focal length F of a typical satellite dish is given by F= D²/ 16d where D is the diameter of the dish and d the depth. So, for a 60cm wide dish we might expect a focal length of about 20cm. If we assume the pickup box to be 5cm across, this means the sine of the 'danger' angle we are seeking becomes ± 10/60 = 0.166 or ± 9 degrees. So the dish must point more than 9 degrees away from the Sun in the sky to be safe.

Taking all the above into account this means the forbidden azimuth settings for the receptor dish extend from 58° to 153° and from 207° to 302°. Large chunks to the east and west are thus excluded. Indeed, in total we have lost more than half the circumference, all because of the Sun and the seasons! The only way to reduce this is to use the smallest possible pickup boxes.

Here is a projected service area map from a notional North London site, assuming reception as far as the optical horizon, and showing the forbidden reception zones as faded areas. It will be desirable for this problem to be overcome if the project is to go forward, and the receptors will have to be made proof against being cooked by up to 250 watts of solar power (for a 60cm dish).

One way through this might be to move the transmitting location northwards - and for simplicity, dispense with the northern 'fan'. The concentration of the energy into a southern fan would increase the efficiency of the system in that direction and would still catch most of London.


 

THE SAFETY ZONE

This is where I shall bring together all my discussions on safety - mainly eye safety at this stage - and set out the safety protocol I shall be adopting in the workshop from now on.

I value my eyesight! I am keen to ensure it is put at no risk, nor the eyesight of others. It would take just one moment of carelessness or absent mindednes to cause a life-changing accident. This is a particular risk because the infra-red light is invisible. Infra-red light could cause a retinal burn without any protective 'blink' response.

Incidentally this is still an amateur endeavour. I am not prepared (at the moment) to fork out an extortionate 210 swiss francs or whatever for copies of IEC 60825 and IEC 62471, which contain abstruse mathematics and lead on to ever more sub-standards, consultants and expensive test laboratories - shrouding the whole subject in an impenetratable fog.

So what about some common sense? All I need to know is one figure: the maximum permissable power into the eye from a point source. The best I have so far been able to glean is the figure of 1mW - for a short time. To put this in context, looking at the Sun with a daylight-adapted eye delivers about 20mW.

I welcome all advice.

I have been wondering some more about the safety aspects of using these point-like LEDs en masse. Thinking of helicopter pilots etc...Initial calculations indicate that a dark adapted pupil would receive 1 milliwatt from the full array at a distance of about 6 metres. This would notionally be as intense as the Sun and while it may be marginally 'safe', it is far too bright. However, at this distance, the radiation would not be coming from a point but rather a distributed source (of many points), so it is not so bad as it seems. Maybe, instead I should be looking at the 'safe' distance from just one of the points - a single LED....

Total flux from 1 LED = 70mW. This diverges out as a fan of 6 degrees. The collection area of a 7mm pupil is 38.5 sq mm. At what distance from the LED will the total flux collected by this area fall below 1mW ? Well, the area of the end of the emerging light cone at this distance will be 70 x 38.5 sq mm = 2695 sq mm. To get the radius-squared of such cone we divide by pi. This is 858. So the radius is 29 mm and the diameter 58mm. Dividing by the 'form factor' (sine or tangent) of a 6 degree cone (0.105), this means the distance is 552mm. So 1 milliwatt will be received by the unadapted eye at 552mm from one of these LEDs. This is invisible light as intense as the Sun so I had better be careful in the workshop! Let's reduce this exposure by a factor of 1000. From the inverse square law, this means the "safe" distance becomes 17½ metres.

So, all-in-all, it looks it will be safe for the 'helicopter pilot' to fly within "several" metres of the radiating array while it's in action. Since s/he is likely to tangle his/her rotor blades on the tower well before this, I doubt the temptation to fly too close will be very strong.

I also need to consider the eyes of birds. Really, I need to design a system that will be safe viewed at any distance.

I have been making good progress with the latest rig, which will have a theoretical peak optical power of 840 mW at 850nM. This is equivalent to an omnidirectional emitter of power: 50 watts. This has the potential to be seriously dangerous when close-up.

INFRA-RED WORKSHOP SAFETY PROTOCOL

  • Never look into the beam - at any distance! Although the energy received from the source will decline with the inverse square of distance, the visual 'intensity' of the ever-smaller source will not.

  • Turn your back to the beam when pointing the receptor and walk backwards returning to the source in the workshop or close your eyes!

  • When working on the bench, the apparatus must remain horizontal at all times, with the emitter facing away from you.

  • During initial setup, a tunnel must be fixed to the front of the apparatus, to prevent accidental viewing of the emitters, even from way off-axis. During experiments, all glass areas in the workshop must be covered up, to prevent the risk of specular reflections.

  • After the apparatus is switched off, or during any interruption - such as from the phone - pull the mains plug from the wall socket. Always use the same socket. The apparatus must never be left running by accident.

  • During experiments, an infrared-sensitive video camera and monitor will be used to render the reflected (off the wall) beam visible. This will also be used to inspect the apparatus generally, to reveal any light leaks.

  • No experiments will take place when anybody else is present.

  • Use infra-red blocking goggles.


So let's look at the current apparatus. The divergence angle will be 6 degrees (±3°@-3dB) and the 'form factor' of this is 0.105. Therefore, assuming a point source (it won't be, but we need to load up the pessimism factor for issues of safety) the beam will have a diameter of 10.5cm at a distance of 1 metre. The diameter of a dark-adapted pupil is 7mm and its area is 38.5 sq mm. At what distance will 38.5 sq mm gather up less than 10uW?

At the 'safe' distance, the total flux of 1 watt will be spread over a disc (perpendicular to the beam) of area: 38.5 x 100,000 sq mm. That's 3.85 million sq mm or 3.85 square metres. Divide by pi to get the radius squared. That's 1.23. The square root of this is 1.1, meaning the diameter of the beam at this distance will be twice this or 2.2 metres. Divide this by 0.105 and we get the distance from the source: 21 metres.

For a full milliwatt into the eye, the distance would be 2 metres. The moral of the story is that this beam will be dangerous in a domestic setting.

I have been having problems with RF oscillation on the new high gain amplifier and blew this up while trying to deal with it. I still had obtained results from the rig, but it was not operating at full efficiency.

The new safety protocol is being rigorously applied, and works well - there really is no alternative! The safety goggles have been tested with an infra-red security camera and have been verified as effective.

I am wondering whether an alternative way forward would be more practical. This would involve using small low power directional units on the transmitting tower rather than a high power 'fan' array. Each unit could be paid for, at modest cost, by the viewer to which it was pointing. For a specialist service with only a few 405-line viewers this might present a workable arrangement. It could allow experience to be gained and would not rule out a fan array later, once the service was established.

Following subsequent discussion of the above on a private internet forum, it has been agreed this is not such a good idea. Although a directional demonstration over several miles might be used during publicity for the system to 'prove that it works', it would be better to provide the 'full fan' at actual launch. Lots more people could then have a go at picking us up, using their own resources.

There is a line-of-sight view from my bedroom window to Snowshill, 7½ km (4½ miles) away. This will provide a useful test path. Later on, Highgate (London) to Welham Green (18km / 11 miles) might be tried, if a line-of-sight path can be established and consent can be gained.

Once I have determined the actual performance in the field of the 1 watt beam, I will know more about what will and won't be possible. The next step will then be to get to work at the receiving end, developing an outfit probably based on a plated satellite dish. This will greatly increase the receiving gain beyond the existing 2-inch lens.

Here's the unit with an attenuator funnel attached to the front... a safety measure.

 

Results using the attenuator funnel. Note the infrared view of what's going on - on the smaller left hand monitor.

I have been considering a change of tack. This is to develop a modular system based around 'modules' each containing clusters of a few LEDs. This could then in principle be scaled up to any size. This has the advantage that the system could grow almost 'organically' step-by-step and there would be no sudden big investment in heavy duty plant later on with the risk of new problems arising.

I am hoping it will be possible to build each module around a single MOSFET driving transistor. Eventually, the modules would plug in to a main frame. The whole system would be wide band and from the start, would be suitable to carry a range of services in addition to 405-line television.

Someone has asked me what the final transmitting array will look like. This will be determined by (1) what sort of structure(s) will be practical to mount on a mast, the existing regulations, windage and weather resistance and (2) what will provide sufficient close-up visual safety (don't forget the birds!) and resistance to long range scintillation in reception.

At the appropriate time I would approach others for advice on (1). As for (2), this would indicate a spread out rather than a compact array.

The main thing now is to develop the plug-in module. It is the design of the feeder frame into which these plug that would later be conditioned by (1).

Also, if I can get to work on the receiving end to greatly increase the pick-up sensitivity (I regularly receive details of the latest technology - apparently there are now such things as semiconductor photomultipliers, though gord knows how linear - or affordable - they are) then (2) could require much less transmitting power and be intrinsically safer.

Some first thoughts (at the transmitting end) on the structure of the FET-based radiating cells. They would probably be square rather than hexagonal, since this is easier to make. They would be made of a white RF-friendly plastic.White would tend to diffuse and preserve stray radiation rather than absorb it. Each cell would be fronted with a weatherproof RF screening panel bearing a hole. The radiating LED would sit a little way down inside a 'tunnel', to protect it from the weather and pollution, to reduce the capacitance from the front panel, and to asssist with visual safety. Using this arrangement would limit the number of LEDs visible from any one position, but only when close up.

To counter scintillation effects when receiving the signal at a distance, it's possible that the transmitting (emitting) array might be constructed in 'triad' form. The energy from such an array would mainly come from its edges. A triad, or bunch of three large LED clusters, would present a relatively large apparent size as seen from any direction. This would mitigate scintillation. Another advantage is that the LEDs would all be a similar distance from the RF feedpoint, reducing the risk of modulation phase cancellation.

The panels bearing the LEDs on a transmitting array would be tilted outwards from the vertical by slightly less than 3 degrees. This would mean the vertical polar diagram of the 6-degree divergence LEDs would beam toward the horizon (but not above) as well as down slightly toward closer locations. The closest location thus served would depend on the height of the array. A spirit level would need to be incorporated in the array for initial setting up.


A prototype module circuit was successfully made. Each module was to power ten 70mW LEDs. The MOSFET transistor was ideal for powering LEDs, since it acts as a current source when in saturation mode. Modulation depth applied to the vision carrier was 90%, though I had little way of knowing how faithfully the LEDs are following this in their actual light output, beyond looking at the received television picture.

The infra-red LEDS have proved somewhat disappointing in this respect, despite being effectively driven. One of them (I masked off the other nine) could only just transmit a noise-free picture across the workshop. I also tried twelve TCLR 5800 red LEDs, leftovers from the Mirror Screw project. These were even worse. It seems that most types of LED, however well you drive them, just will not modulate fully at 45MHz. The best results I have had so far have been from the very expensive special LEDs, specifically stated to be good for 70MHz. One of these was used at the Loughborough demonstration.

I now will have to take a step back and take a fresh look at this project. How can I produce a really powerful beam, modulated at 45MHz ? Is it worth looking at a laser system again? The saga continues.


Another change of tack! To cut a long story short, I have not been impressed with the modulation performance of the infrared LEDs I have been using. According to my scope, I have been current-driving them with a signal of good modulation depth. Although the quality of the received picture has been good, the reception range has fallen below that expected. The reason must be that these LEDs are only sluggishly following the applied carrier signal.

Having said that, they work far better than some standard red hyper-bright LEDs I have tried. The only LEDs that worked really well were the expensive 'resonant cavity' LEDs, which were specifically rated for upto 70MHz.

It seems that, even if I feed the device's capacitance from a low impedance to mitigate its effect, the LED itself must also be intended for the job.

The inverse square law is not my friend! One problem I now have is that the number of LEDs required at the transmitting end to provide sufficient power for the current receptor - and dish - at say 10 km - would be prohibitive. It would also be extremely boring wiring them up along with their amps, while I added more and more stray capacitance!

Moreover, the whole assembly would be bulky, hard to mount and make weatherproof, and contain relatively complex electronics.

So I am now looking into another way forward.

This would involve using a series of modulated lasers at the transmitting end. This would be so much lighter in weight, easier to RF screen, generally simpler and more elegant! The beams would overlap side by side to reduce scintillation effects. Each would emit a defocussed horizontal 'line' or flattened elliptical beam pattern. They would be visible red to facilitate aiming, but of only modest power for eye safety.


I believe I would need much more sensitivity at the receiving end than the current prototype receptor can supply. To quantify this, I am now carrying out more calculations and swotting up on the latest sensors, notably the APD (avalanche photodiode) and the MPPC (multi-pixel photon counter) or SiPM (silicon photomultiplier). As usual, I am pushing things - with the need for linear response and a VHF modulation frequency.

It yet remains to be seen if I can "square this circle". If you have any experience or tips regarding the above devices, please get in touch.


The plan now is as follows:

  • AT THE RECEIVING END, to obtain a suitable silicon-based avalanche photodiode (APD) module, complete with integral power supplies, temperature compensation and amplifier - for trial. I have so far identified two candidates. This should increase receptor sensitivity.

  • To mount the above, complete with local collector lens, on a chromium plated satellite dish. This should still further increase receptor sensitivity.

  • AT THE TRANSMITTING END, to obtain a suitable wide-band modulated laser for trial. The type, beam profile and power of the laser will depend on the flux requirements of the new receptor, taking into account visual safety. The laser is likely to be used in (blurred) horizontal line mode and several of them would probably need to sit side by side on a future tower.

This project was originally conceived as a platform for a non-commercial application - 'heritage television' - using an obsolete technical standard and mainly of interest to committed enthusiasts. As a result, the intrinsic unreliabilty of light transmission due to weather was seen as less of an issue.

This rather ruled out a commercial data carrying application. However, the method still has its attractions, since it would be able to transfer large amounts of data periodically, one way, without any fibre optic infrastructure needing to be in place. This could mean a highly competitive price.


I now pause whilst I again consider the following questions:

  • Will it work? Can the sensitivity at the receiving end be made sufficient to work from an eye-safe output at the transmitting end?

  • Do the required modulated laser products even exist? This is ultra-wide band modulation, remember.

  • Will its development be affordable? Have I enough money that I am prepared to risk?

  • Will it eventually pay? Is there a commercial application?

LATEST NEWS

I've been thinking about matters at the receiving end. Maybe a chromium plated satellite dish will be a mite expensive, and even though its f-ratio will be short, could be a fiddle to aim and set up. Is there any alternative that can gather a lot of light while being non-critical to aim?

Well, I understand a new technique using a 'bulb' of fluorescent optical fibres is now in the pipeline. The fibres catch the light over a broad area, fluorescing as a result. The fluoresced light is then piped down to a small collection point. The advantage is that this method combines non-critical pointing of the array with high speed detection. Trouble is, since it depends on fluorescent effects it has to use the most energetic - blue - light. We're using red here, to penetrate the mist. Anyway, I can no longer find the link to this!

Another idea for a very low f-number (easy to point) arrangement would be an exponential horn. However, the light collection ability would still be set by the size of the mouth of the horn.

Still another idea would be a 'squarial' - a flat panel consisting of innumerable little low capacitance sensors plus combining electronics. This, I imagine, would be a 'big bucks' solution well beyond my capacity as a 'small guy' to finance or develop.