[RobRusbridge]

In the early 1990s I was part of a team from the Optoelectronics Research Centre at the University of Southampton who did a demonstration at (I think) the Royal College of Physicians in London when we won the Rank prize for our Optical Amplifier which directly amplified optical signals in a length of fluorescent fibre where the power was provided by a fat-ish infra-red laser, so no need to detect to electrical signals, amplify, and re-modulate. We had a source laser diode set up in a "bias tee" network sending a beam down a few kilometres of fibre, wound up on drums, through an Optical Amplifier, a few more kilometres of fibre, and all detected on a simple but quite fast photodiode with a simple resistive load (we had bags of light signal after all). We hooked up a pile of then-current VCRs, all tuned to put their RF carriers at different frequencies, and for good measure used a set-top indoor aerial to add the terrestrial (then analogue) channels into the mix of one of the VCR's outputs, and fed the whole "signal soup" to the laser diode, each signal source having its own feed resistor into the bias tee. At the receive end we had several TVs with at least twelve channels of push-buttons on the tuner, and every one was tuned up for all 12 different signals. We fed the recovered signal soup to the aerial socket on the TVs through a little aerial distributor box and we could flick channels beautifully. Four different channels, eight different videos, receive-end quality easily good enough to clearly see the limitations of a VHS tape compared to a proper TV channel. I seem to remember we had Top Gun on the TVs while the boss-man was talking to the visiting VIP - one Margaret Thatcher. You may remember her.

I have no doubt that a modern fast LED will do what you want it to over a reasonable distance. The physical vibrations are a problem in line-of-sight communication across a room, but even so, you should be OK. You can modulate an LED with a bias tee just like a laser. You run the laser or LED from the current-limited supply through a resistor in the positive feed (as long as the current source doesn't panic about having to shove out a volt more than it's expecting) and add one or more other resistors feeding to the anode of the diode, where it joins the feed resistor. You only have to couple in the signal(s) to the remote end of the resistor(s) and the small fluctuations of current caused by the signal through the feed resistor modulate the brightness of laser or LED because the current source will be stabilising current at low audio frequencies and it won't have anything to say about the sort of frequencies you'll be modulating on there. You want to be pretty sure you're only going to get signals of the amplitude you're expecting, though. A big transient will pop the laser. Worn that tee-shirt. At the time, a decent laser diode packaged up pre-aligned to an optical fibre was about £1500. Er, oops. It came off in my hand, officer.

(On the way home, all the kit was piled into the back of a heavily loaded Peugeot 405 estate. As Janet spotted the amber traffic light make an appearance beneath the red, she gave the right pedal a press and the left pedal some upward movement, the boot popped open and half the contents of the boot emptied themselves right across the junction. I was in my car behind. All I could do was sit tight at the stop line, switch on the hazard lights and try not to laugh enough to actually burst a blood vessel. Dancing around in the junction, piling things back into the boot, were Janet, who did most of the work on the fluorescent glass fibre, and Richard, who had done most of the rest of the work. A unique pair of academic stars trying to avoid the oncoming busses.)