481THz over-the-horizon tests

Non line-of-sight
(NLOS), or over-the-visible horizon, is a facet of optical communications that I find particularly fascinating: even though there is no visible sign of the red light from the TX in the sky at the receiving end, it is possible to detect narrowband signals such as QRSS3 at considerable distance over the horizon using cloudbounce (bouncing off the base of clouds) or clear air scattering (signals forward scattered by dust and water droplets in the lower atmosphere).

My recent tests (13.3.13 and 18.3.13)  have been with clear air scattering at night over distances of 4.8km and 8.5km over a path with intervening rising ground that fully obstructs the line-of-sight from my home QTH to a remote receiving sites on a roadside. My TX beacon is a 1W 10mm diameter red LED running at 250mA in 100mm optics firing out through the glass of a double glazed bedroom window. The RX is now an SFH213 PIN photodiode in a K3PGP based design in 100mm optics which is very sensitive at low audio subcarrier frequencies in darkness. This receiver is easily degraded by any light. Alignment is VERY critical and just a few degrees out results in over 20dB reduction in S/N. I am finding the tripod us
ed at the RX end (a cheap camera tripod) is too flimsy to really use seriously: it is hard to align repeatably and I need a much better, stronger tripod with a way of resetting directions and elevations to with 1 degree.

My best results so far are 14dB S/N QRSS3 signals in 0.34Hz at 8.5km and 20dB S/N signals in QRSS3 in a 0.34Hz bandwidth over a 4.8km path. With signals this strong (as long as the RX and TX are VERY accurately aligned), it is believed that much greater distances should be possible NLOS by clear air forward scattering. I suppose you could call this optical frequency troposcatter.  Ranges by cloudbounce have yet to be ascertained. 

To get best results with clear air scattering it seems that the TX and RX have to be aligned just above the horizon. Elevating the RX results in a dramatic drop in signal level. Much more work is needed to optimise the elevation angles with other NLOS propagation modes. For cloudbounce, the optimium elevations over short paths are likely to be much greater than for clear air scattering.
Both are best played back using Spectran software with 0.34Hz bandwidth settings. The signal is inaudible by ear and there is no sign at all of the red beam visible in the sky at the receiving sites, yet in a 0.34Hz bandwidth they are good solid signals.
The 8.5km non line-of-sight path successfully spanned on 18.3.13

14dB S/N signal in 0.34Hz bandwidth over 8.5km

The image shows how strong the signal was at 4.8km - 20dB S/N