Comments on using subcarrier systems on optical voice circuits using LED-based transmitters

Baseband versus subcarrier:


In our experiments, we have used "basedband" modulation for the conveying of voice modulation atop a modulated light beam and with it, we have managed to break a number of distance records.  In a nutshell, this system may be described thusly:

The above system is very simple, requiring no exotic equipment other than the easily-built modulator and detector.  Being "baseband", its frequency response need only range from 200 Hz or so to 2500-3000 Hz to convey speech and this spectral range is low enough that the inevitable degradation due to the capacitances of the detector's active devices (e.g. photodiode, input amplifier, etc.) and the operating impedance is reduced by the maximum degree practical.  Because of the nature of the system, its ultimate sensitivity is unmatched by any other for voice-grade communications.

Subcarriers - FM

Subcarrier operation implies that the voice information is conveyed at a frequency higher than that encountered in baseband.  In past years, many "optical communicator" circuits have been described that utilize a frequency-modulated (FM) subcarrier at a fairly high ultrasonic frequency - typically in the 30-80 kHz range.  Because it is FM, its occupied bandwidth is, when modulated by speech, in the 10-15 kHz region, hence the need for the fairly high frequencies.  FM has the obvious advantage in that it is, by definition, immune to amplitude variations in the carrier level, the intelligence being conveyed solely by the variation in frequency, and as such it may be minimally affected by amplitude fluctuations due to scintillation, insects, rain and interference from lighting - provided that the minimum signal/noise level is above the threshold of the FM detector!

When using an FM subcarrier system there are several things that can contribute to its lesser ability to be used under weak signal conditions than straight baseband operation:
(2 * 2.5 kHz deviation) + (2 * 3 kHz modulated frequency) = 11 kHz.

If this is compared with that used by baseband - which is simply that of the highest audio frequency (e.g. 3 kHz) one can see that nearly 4 times the bandwidth is required - and communications theory tells us that the wider the bandwidth in front of the "demodulator", the more "diluted" the desired signal becomes from the various noise sources.  In other words, we may get more noise in our 11 kHz FM detector bandwidth than we do in our 3 kHz baseband bandwidth!

In the simpler circuits, this detection bandwidth is often ignored:  Many FM "optical communicator" schemes consist of a simple voice-modulated oscillator on the transmit end and a PLL-type detector on the receive end - but there is often NO attempt made to precede the detector with a bandpass filter and more often than not, the output from the optical receiver is simply dumped into the detector unfiltered - noise and all!  To be sure, a PLL-type of detector does, in fact, have a somewhat limited bandwidth intrinsically, but even off-frequency noise can degrade its performance owing to the necessity of such a detector to have limiting action.

The "better" FM subcarrier systems will, in fact, have a proper bandpass filter preceding the FM demodulator - either an op-amp type filter at the ultrasonic frequency of the subcarrier itself (say, +/- 5.5 kHz from the center of the subcarrier) or that ultrasonic frequency may actually be up-converted to a higher frequency and applied to an already-existing FM communications receiver:  Past articles have appeared that convert the ultrasonic frequencies to HF or even VHF to allow the FM mode of a radio (or handie-talkie) to be used - proper filtering and all!  In addition to bandpass filtering, a narrowband FM scheme will also employ pre-emphasis on transmit and de-emphasis on receiver - a clever (but simple) trick to further-enhance the performance of an FM system under weak-signal conditions.

Provided that the detection bandwidth is taken care of, there's still the issue of the FM threshold.  While FM is generally immune to variations in amplitude causing effects on the demodulated signal, that signal must maintain a minimum signal level above the noise in order for degradation due to that noise to be avoided.  While it varies with the FM demodulator method and bandwidth, this threshold is generally in the 8-10dB range for a fairly "noise-free" signal and below this, the recovered signal rapidly gets lost in the noise!

In baseband operation, the skilled listener can easily make out speech that has just a 6 dB of signal-noise ratio - and this is not only 2-4 dB lower than that detectable by an FM-type circuit, but because the bandwidth of the baseband is less than 1/3rd, it's also less-affected by the noise as well!


Finally, there's also another issue that arises in a practical subcarrier scheme using common photodiode detectors:  The loss of ultimate sensitivity as the frequency is increased.

As noted above, the capacitance of the photodiode combined with the high impedance of a very sensitive detector conspires to form a low-pass filter.  For practical photodiodes (e.g. 1mm2 to 10mm2 in area) and typical circuits this "knee" frequency (where the frequency rolloff starts to become apparent) typically occurs right in the voice range of 200 to 2000 Hz, depending on the particular circuit and photodiode used.

Above this "knee" frequency, the audio response of the circuit drops off at approximately 6dB/octave while the noise floor itself generally stays constant (once one gets above 1/F noise - which we will ignore for the purposes of this discussion) which means that the signal-noise ratio (and ultimate sensitivity) of the detector as a whole decreases with increasing frequency.  Knowing this, one quickly realizes that for purposes of ultimate system sensitivity, it is best to use as low a frequency as possible!

If we take a 50 kHz FM subcarrier as an example using a receiver with its "knee" at 1 kHz, we can quickly calculate that at a rate of 6dB/octave, we may lose 30-40dB of ultimate sensitivity - this, atop the degradation caused by our necessarily-wider bandwidth and the lower threshold of our FM detector!  Practically speaking, it's been observed that a well-designed receive system can, to a degree, mitigate this degradation, but even the best implementation will suffer somewhat!

Why use FM then?  The main attraction of FM is, again, the fact that it's insensitive to amplitude variations - and provided that the minimum signal level is maintained, it can be essentially noise-free:  If the path is fairly short, the air is clear and the transmit/receive optics are reasonably well-designed, this scheme can work very well.

The avoidance of man-made optical interference:

An advantage of subcarrier schemes not yet mentioned is that the necessarily-higher frequencies involved move it away from the "buzz" and "hum" of manmade urban lighting which contains strong components of twice the mains frequency and its harmonics (e.g. 100, 200, 300, 400 etc. Hz for 50 Hz mains and 120, 240, 360, 480 Hz for 60 Hz mains.)  If the optical path involves passing over/near areas in which such lighting is used, the baseband frequency range can become heavily contaminated with the mains-related harmonics, possibly making communications difficult.


These mains-related harmonics (buzz, hum) may be filtered out with the proper equipment:  For more information about how this may be done, read the page "A Comb Filter to combat mains-induced hum from urban lighting" at this web site.

Fortunately, the amplitude of these harmonics tends to drop off fairly rapidly with increasing frequency and by the time one gets to 10-20 kHz, they are typically reduced to the point of either being inaudible or minimally intrusive and in this way interference from mains-powered lighting may be avoided.


The use of SSB subcarrier schemes:

More recently (2010 and later) the UK optical group has been a proponent of the use of single sideband (SSB) on optical system - and for some valid reasons:

To be sure, the use of SSB is a bit more complex than simple baseband - and from a purely theoretical standpoint its weak-signal performance is going to be demonstrably inferior, but based on practicalities and the environment experienced by the UK folks it does make some sense:

In the future, I hope to add more links, but in the meantime one may join and peruse the UKNanowave Yahoo group for more information.

If you have questions or comments concerning the contents of this page, if have information about other suppliers, feel free to contact me using the information at this URL.
Keywords:  Lightbeam communications, light beam, lightbeam, laser beam, modulated light, optical communications, through-the-air optical communications, FSO communications, Free-Space Optical communications, LED communications, laser communications, LED, laser, light-emitting diode, lens, fresnel, fresnel lens, photodiode, photomultiplier, PMT, phototransistor, laser tube, laser diode, high power LED, luxeon, cree, phlatlight, lumileds, modulator, detector
This page and contents copyright 2007-2012 by Clint Turner, KA7OEI.  Last update:  20120725