Monday 29 December 2014

VK4EBP's NLOS Lightwave Experiments - Part 2

Once again, from Jan...

"Visible LEDs and optics: Some of my favourites include the LZ1 and LZ4 series from LedEngin used with matching spot lenses, or thick condenser lenses from old visual projectors. Also various Golden Dragon varieties with matching small plastic spot lenses. Average beam divergence with either configuration is about +-4 degrees. (Happy to provide detailed parts numbers.)

Infrared LEDs: Here we have a some good products from
Vishay (TSHG series), with narrow beam (some go down to +-2 degrees) in the standard 5mm packages and max current of 100mA.

They can make nice pencil-beam arrays in combined series/parallel with 200 or more LEDS and no optics required, at less than 40c per LED. Lots of soldering though :). For the 950nm range (not extensively tested yet) the Vishay TSAL range of LEDs will do nicely.

LED drivers: I found a very neat IC to PWM-drive the LEDs at up to 1A - the CAT4101 from ON Semiconductors.

Usually have two in parallel for a peak of 2A at 50% duty cycle, giving me 1A average to drive the LED string.

At the reciving end I am now using a preamp loosely based on
Clint's no-feedback design with some hum filtering and other post minor post-processing feeding to an audio amp, or to SpectrumLab."
"...tonight I am trying to compare 830nm against 940nm on a 600m short NLOS path. For this, I have just built a small TX with two "naked" LEDs - TSAL6100 (940nm) and TSHG8200 (830nm). Both have identical shapes of their radiation patterns with +-10 degrees. Each mounted inside of a square cross-section black cardboard tube side by side to ultimately give them a very uniform square beam of approx +-4 degrees. Each driven with 440Hz square wave at 100mA with unique morse identifier for each.

There is a third visible LED as well with its own cardboard collimator and morse ID, to serve as aiming tool and a kind of reference.

The purpose is to see whether 940nm would offer additional advantages in NLOS situations, due perhaps to better/different reflection, scatter, refraction or whatever physical processes might be involved.

An additional advantage might be removing my setup further away from visible light pollution. With the growing popularity of LED lighting in both household and the industry we are having a chance of less and less infrared pollution from traditional incandescent or similar sources - and making light comms more practical in urban environments - all with a simple IR filter at the receiving end.

Tonight's tests were not as productive as anticipated, but somewhat educational nevertheless. First of all the receiving location I picked on the map suffered from severe QRM from sodium street lights. I picked strongly all my "big" IR transmitters, but the visible and the "naked" ones were lost in the QRM, if present at all. I returned home, pointed the 150mW TX into nearby bushland and went for a walk. The red light reception vanished immediately after loosing sight of TX, followed quickly by the 940nm with the 830nm persisting the longest up to perhaps 100m - with the beam fired parallel to ground from upper storey into the crowns of the trees. Reception required pointing the receiver slightly up, intersecting the TX beam somewhere in the tree branches.

Briefly - I confirmed what I already knew:
- IR is far superior to visible for NLOS work - indeed visible light is of no use.
- Low elevations and scatter from ground objects is superior to high beam elevations.

A new observation (that I would like to confirm) is that 900+nm is not worth bothering about. It is known that 950nm suffers from greater scatter in atmospheric particles than 850, but this proved to be more of a hindrance than help in NLOS work.

Another practical observation is that the 950nm version of the common SFH213 photodiode (SFH213-FA) works very well in receiving 850nm whilst filtering out lots of the visible pollution (well, perhaps except the sodium lamps!).
I started tonight's session before sunset and got very good NLOS reception of my large IR TXs - in what could be described as quite a bright twilight. (I carry two plug-in front-end modules each with one of the two photodiodes and the input FET.)
Another holiday project - a pocket-sized 10W TX! I found new IR LEDs - SFH4783 - rated at 2W, barely 1.65V of Vf, and intrinsically narrow angle of +-10 degrees. This means no optics, and up to seven of them in series on a small heatsink can be run from a 12V SLA battery.

Osram SFH4783

This reminds me of yet another observation - out of my several large 850nm TXs the best performers are the naked narrow-angle multiple LED arrays - and the one containing 3 high power broad-beam LEDs with spot optics performing the worst. Well the lenses are designed for visible LEDs and I have no guarantee that the material refraction angles and loses are acceptable in the IR range... 

Returning to Steve's questions and our general discourse: it (the RX) fairly small and portable...and lensless?   (cont'd)...


Unknown said...

"required, at less than 40c per LED. Lots of soldering "

Where can you get these at 40 cents each?


Steve McDonald said...

Paul...the commentary on this three-part blog is from Jan so I'm not sure where he sources his components. I do however, see his Vishay TSAL6100 IR diode selling at Digikey for 42 cents each for 10 ....even cheaper if you buy larger quantities.