Take a look at this website when you get time.

As far as I can tell, they used the BPW43 sensor with a simple Bi-Fet buffer directly into a differential video amplifier. They were working with a 10Mbps signal, so they needed the video amplifier, but the front end would work just as well for our audio signals. There does NOT appear to be any optical filter in front of the BPW43 sensor. They used either a 625nm Red LED or an 875nm IR LED as an option.

This is their sensor mounting scheme (from https://ronja.twibright.com/receiver/building.php )

The box with the LED protruding contains all of the circuitry for the 10 Mbps link and as far as I can tell, the whole box is mounted inside their "stove pipe" tube painted flat black on the inside. They put the lens at the front of the tube and a long light shade beyond that. It appears to be that simple and yet they are getting a 10 Mbps data rate with no errors at 1.4 km with a 5 inch 2X magnifying glass for a lens.

Their transmitter is a single LED that is less powerful than what we are using now (one led from the tail-light of a car from the period - 20 years ago) and the same size 5 inch 2X lens.

It is interesting that they put  4 resistors on the lens to prevent fogging or condensation.

That is pretty encouraging. With my single LED flashlight (no fresnel lens), and my  KA7OEI receiver with the fresnel lens and a short light hood on the sensor, I was able to go 635 meters with a 30 dB margin in full daylight. As I explained, the flashlight was against a wall of trees in the shadow on the north side of the trees, and the field of view of the receiver only saw the flashlight and the trees. So the background and field of view matter. But the Ronja unit appears to work without much concern for the background.

I am looking forward to working on a "Daylight" version of our system. If not "Daylight" then at least "Golden Hour" or "Twilight" capable. More experimenters welcome :-)

Rob - K0XL / K0THZ