How the fibre communications revolution began

Discovery of the problem:

In 1977, STL/STC installed the world’s first field demonstration of a fully regenerated optical transmission system between Hitchin and Stevenage. This 140 Megabits/second system used Multi-Mode fibre and quasi-single longitudinal mode semiconductor lasers. When the system was set up in the laboratories prior to installation, high levels of noise and distortion were observed on the received waveforms and it was evident that a potentially serious problem with fibre systems had been encountered. Although first attributed to “bad” lasers, it soon became clear that the problem was of a more complex nature. Most of the six regenerators exhibited the effect, and paradoxically, those lasers which were otherwise considered “good”, produced worse “Modal Noise” at the following receiver. Subsequent experiments showed it to be a complex system problem, caused by an interaction between several components (and strongly dependant on some specific parameters in each case):

1/. The source (spectral purity, temperature, reflections, modulation).

2/. The fibre (number of modes, bend condition, temperature).

3/. Fibre to fibre joints (offset, losses).

The mechanism:

I christened the effect “Modal Noise” very early on, because I had a hunch that it had something to do with the fibre modes. After a great deal of painstaking experiments and studies, I developed a complex model which satisfactorily explained all the effects we observed. This work was finally published in an invited paper at the 4th European Conference on Optical Communications, 1978, Genoa, Italy, “The Phenomenon of Modal Noise in Analogue and Digital Optical Fibre Systems”, [1].

The simplified concept is as follows:

The high temporal coherence of the lasers produces speckle patterns within the fibre cross section. Changes in these speckle patterns are caused both by bending the fibre (which changes the phase delays between modes), and by fluctuations in the optical frequency of the laser (caused by the modulating the laser and by reflection instabilities). These changes are harmless in themselves; however Modal Noise is “generated” at a discontinuity such as a misaligned joint because only a part of that speckle pattern is transmitted past the joint. The fraction transmitted is modulated by the changes in the speckle pattern, because the brighter speckles may or may not be coupled to the following fibre.

Initial resistance to the idea:

The problem only became conspicuous to us when a real system was constructed, with several repeaters. Before publishing my findings I asked many people working on multimode fibre systems in the UK and abroad, if they had observed any unusual received signals. All denied seeing any problems at the time, the competitive climate and the need to sell the idea of fibre systems to potential customers, made other research teams blind to these phenomena.  If changing the laser made the problem go away, it was forgotten, (a defective laser almost always has lower coherence). Some of my own colleagues refused to believe the evidence and my theory, simply because no one else had reported such effects. One of them, to support his argument showed me a Japanese paper which purported to give excellent results in a configuration which my theory predicted would be catastrophic. I later found that although the figures in the paper showed reels of fibre, subsequent discussion with the author revealed that the experiments had actually been performed using a simple bulk optical attenuator to simulate the fibre. The author had not realised that the difference was significant, and that he would be found out!  The paradox is that despite the initial resistance, once Modal Noise was accepted, it was so easily observed.  Though sometimes other phenomena were confused with Modal Noise, as was the case when the noise was due to laser reflections instead.

The solution:

Early investigations indicated that laser coherence was at least one cause of the problem, so we urgently experimented with ways of reducing the laser coherence to find a quick solution for the Hitchin-Stevenage link. The solution in this case was to change from Non-Return-to-Zero modulation, to Return-to-Zero modulation, allowing the laser to be modulated with short pulses from below threshold. This ensured that the laser could never settle down into coherent single longitudinal mode oscillation. In 1981 I published an article [2], which reviewed the full nature of the problem in multi-mode and single-mode fibre systems, and described how to avoid it. (I have since published many other papers on various aspects of modal noise and modal distortion).

Modal Noise is minimised by using a light source whose optical bandwidth is much greater than that of both the fibre and the information. This precludes the use of most lasers. (It also is an unsatisfactory solution when Chromatic Dispersion dominates).  The only complete solution to Modal Noise is to use single-mode fibre.

The impact on fibre systems:

The discovery of Modal Noise and Modal Distortion (waveform distortion due to Modal Noise), virtually eliminated all hope of using simple analogue modulation of lasers in multi-mode fibre systems, especially in the early days when connectors were lossy due to misalignment.

Modal Noise has non stationary statistics, like the weather or waves at sea. The discovery of Modal Noise revealed that systems using coherent lasers and multi-mode fibre with mode selective loss (e.g. joints, couplers), could not be guaranteed to function correctly, although much could be done to reduce the probability of system failure. In addition we were just beginning to realise that one could not usefully predict the total bandwidth of concatenated lengths of graded index multimode fibre. These were unpalatable truths for those who wished to replace co-axial copper cable (which is single moded and entirely predictable), with multimode optical fibre which is not. The discovery of Modal Noise was the primary reason why we pushed for the early adoption of Single-Mode fibre in the UK network. Within 2 years, attention was turning to transoceanic undersea system development. It was obvious by then that multimode dispersion would severely limit the performance of any transoceanic system. The 1980 deep water cable/repeater trial in Lock Fyne incorporated a single mode fibre.

So finally, Charles Kao’s 1966 vision of a true single mode high capacity transoceanic fibre system became a reality.

These problems forced us to acquire a good understanding of coherence in lasers and systems [3], and consequently gave us an early start in the development of coherent optical fibre systems.

Richard Epworth – September 29, 2017

References:

1              “The Phenomenon of Modal Noise in Analogue and Digital Optical Fibre Systems”, by R.E.Epworth, Proc. of 4th European Conference on Optical Communications, Sept. 1978, Genoa, Italy, pp 492-501.

2              “Modal Noise – Causes and Cures”, Laser Focus, by R.E.Epworth, Sept 1981, pp 109-115.

3              “The Measurement of Static and Dynamic Coherence Phenomena using a Michelson Interferometer”, by R.E.Epworth, Proc. of 5th European Conference on Optical Communications, Sept. 1979, Amsterdam, Holland, paper 4.2.