Nuts and Volts - September 2017

September 2017 9

Before we get to the detector, let’s think about what kind of modulation might be desirable. The first order approach would be to switch the laser on and off with each bit that you want to send. While that will work in some circumstances, it may not be robust in a noisy environment.

At the very least, we might want to add some forward error correction, or FEC. We might also want to consider using a carrier signal that would help us to distinguish our signal from other background noise. In a case like this, probably a Manchester coding scheme would work well.

Manchester coding is similar to phase modulation in that there is essentially a 180˚ phase shift between a 1 and a 0. The advantage of this scheme is that a clock can be recovered from the encoded signal regardless of the data in the encoding. That clock frequency can help us distinguish signal from noise.

The principle is quite simple: A 1 is encoded as a 01 sequence, and a 0 is encoded as a 10 sequence. Figure 1 shows waveforms that illustrate how it works.

With this scheme, we have a viable data stream. We can then add FEC-like Reed-Solomon error correction coding to attempt to fix up or at least detect the inevitable errors in the bitstream.

There are many different Reed-Solomon polynomials to choose from which trade off redundancy for resilience. It might require some experimentation to see what the statistical properties of the errors are in your environment, but some Reed-Solomon coding will almost certainly be better than none, regardless of the polynomial.

There is a good article on Wikipedia about Reed-Solomon at

https://en.wikipedia.org/wiki/Reed–Solomon_error_ correction.

There are some interesting choices for detector devices. The simplest would be a photodiode. In general, silicon photodiodes have low noise characteristics, but there are other materials that trade off properties like speed, light spectral response, sensitivity, dark current, etc.

To go to higher sensitivity, an avalanche photodiode could be used. These are highly sensitive, and ambient light filtering becomes critical so that the device is not saturated.

These devices function a bit like a photomultiplier tube (very interesting devices in themselves) in that one photon is effectively multiplied when it strikes a target, and releases more photons by the photoelectric effect that are subsequently detected by a semiconductor junction.

These devices can be very fast. You will need a several hundred volt power supply to bias them properly. I found a company called Laser Components that has an entire line of APDs that might work ( www.lasercomponents.com/us/ photodiodes/avalanche-photodiodes).