History of the Atlantic Cable & Undersea Communications
Undersea Cable Systems - A Survey
Originally published in the early 1970s in the IEEE Communications Society Newsletter.
Sadly, Bob Easton died some years ago; thanks are due to Jerry Hayes for providing the text and pictures, and for obtaining Paula Easton's permission to reproduce the article. See also Leo Parrish's CS Long Lines page, which has photographs of Bob Easton at work.
Hoping to catch her husband before his departure, the Philadelphia housewife dialled the number hurriedly. "Is this Joe's Barber Shop?" she asked the voice at the other end. The unexpected response was, "Lady, this is the Cable Ship LONG LINES, 300 miles off the coast of England." "Oh, for Pete's sake, a wrong number is bad enough, but with my luck I end up with a drunk or a nut," she announced disgustedly as she hung up to try again.
The author, who was the other participant in this little drama that took place a number of years ago, sometimes still wonders if she would have felt less piqued if she had realized that her 100 call had really gone 3500 miles - under 3 miles of water, and had then traveled in innumerable circles of less than 55 feet in diameter for another 300 miles to arrive and brighten the day for one weary, slightly seasick, engineer. For she had stumbled into the middle of the installation of a new submarine cable across the Atlantic. During such an installation, the cable is layed powered to permit continuous measurements which verify that the equipment being consigned to the depths is working properly-and to provide data on which to base equalization of the transmission path. Furthermore, to make any required technical information or expertise available to those on ship-board, a few order wire channels are equipped and, for the duration of the lay, connected to the domestic dial network with an assigned number at an exchange in the vicinity of the cable terminal.
Approximately 100 years prior to the wayward call to Joe's Barber Shop, the first successful trans-Atlantic telegraph cable was installed by The Great Eastern with Lord Kelvin on board as transmission engineer (see Fig. 1). This cable consisted of one long electrical conductor, insulated from the sea by gutta-percha and given strength by steel armor wires surrounding the outside. The cable was tested uninterruptedly to insure dc continuity back to shore and that adequate insulation from sea ground was maintained as the cable was put overboard.
In 1955-1956 the first trans-Atlantic telephone cable was installed by the Bell System. Unlike the passive telegraph cables that preceded it, this cable had repeaters spaced at 38-nmi intervals to compensate for the attenuation of the cable. The initial capacity of this first trans-Atlantic cable was 36 channels, later increased to 48 channels by means of high-efficiency channel bank equipment which reduced the required guardbands between adjacent. channels. This first trans-Atlantic submarine cable system used one cable for each direction of transmission. During laying, measurements were made from shore to ship with communications in the reverse direction via radio. The cable itself was not terribly different in structure from the telegraph cable, and even repeaters were long flexible "sausages" that could go around the sheaves of the existing cable ships-all of which had been designed to lay unrepeatered telegraph cable. That initial submarine telephone system (as well as others of the same design) is still in service, and has experienced only a single repeater failure in its almost 20-year life.
Because it is impossible to avoid intensively fished waters over the relatively wide continental shelves at the two ends, trans-Atlantic cables have averaged about one cable break per cable per year - mostly due to trawlers. (This is an almost negligible problem in the Pacific and Caribbean which have much narrower shelves and where fishing grounds are more avoidable.) Burial, by means of an underwater plow, of cable likely to be subjected to trawler damage has alleviated this problem, and further improvements in plowing currently under development will hopefully make such breaks an increasingly rare occurrence.
Since that initial submarine cable design, the SB system, first installed 20 years ago, there have been three subsequent designs, each one with a channel capacity many times greater than that of its predecessor. Table I summarizes the history of Bell System Submarine Cable Systems. The development of the most recent SG design is just being completed, with the first installation scheduled to begin this year for service in 1976. In order to achieve the tremendous increases in bandwidth and channel capacity indicated, there has been a massive evolution in the physical and electrical configuration of submarine cable systems - which in turn have required new laying and testing techniques.
Today's systems have little similarity with their simple telegraphic predecessors. Modern submarine cables use an armorless coaxial cable design, with the strength provided by a steel strand inside the copper-sheathed center conductor. The dielectric is solid polyethylene surrounded by a copper outer conductor and an outer protecting polyethylene sheath (see Fig. 2). The repeaters have a rigid housing and directional filters which permit transmitting in both directions through one cable and a single amplifier in each repeater (see Figs. 3 and 4). The lower portion of the transmission band goes in one direction, the higher portion in the other direction, with an unused guard-band between the two. Each repeater contains a crystal oscillator with a unique frequency, which is used for fault location and to determine gains and losses along the length of the system.
As indicated by Table 1, successive systems have been designed with large increases in channel capacity which requires large increases in top frequency. Since the loss of a coaxial cable increases with frequency, the total loss that has to be compensated by the repeaters has gone up significantly from one generation to the next, despite increases in the cable diameter which partially offset the increase with frequency (the loss of a coaxial cable in decibels is approximately proportional to the square root of frequency and inversely proportional to the diameter). As a consequence, the spacing between repeaters has gone down by about a factor of 2 from one generation to the next.
In the case of the most recent SG system, the total loss to be compensated is about 30 000 dB. Thus, if it were not for the repeaters, a signal at the output of the cable would have a power 10^-3000 of its value at the input. (It is interesting to compare this to the estimated ratio of the radius of an atomic nucleus to that of the universe, which is about 10^-42.) To compensate for this tremendous loss, the SG system contains about 750 repeaters spaced at 5.1 nmi. Those repeaters are powered in series, through the center conductor, from the two ends. Ideally, the gain of each repeater would exactly compensate, at each frequency, for the loss of the associated cable. In fact, there will be small differences between average repeater gain and average cable loss, as well as variations from repeater to repeater and cable section to cable section. Both of these types of variations lead to deviations from the actual zero net gain of a cable and its associated repeater. Any such differences between these gains and losses have a signal-to-noise penalty associated with them, which is roughly a function of the total net gains and losses between any two repeaters. To hold these penalties to an acceptable level (a few decibels, ocean block equalizers are inserted at regular intervals along the system (every 20 or 30 repeaters). These equalizers are passive; they use gain from an adjacent repeater which has much less than a full cable section associated with it (in the case of SG, 1 nmi rather than 5 nmi). These equalizers contain fixed networks to compensate for deviations known at the time the equalizers are manufactured. They also contain a family of networks that can be switched in or out just prior to the laying of each equalizer to compensate for deviations that first become apparent, as the system is layed. It is the determination and setting to the optimum choice among the available networks for each equalizer that is the most important transmission function during laying.
In addition to the undersea portion of a submarine cable system, there are two terminals. These provide the power for the repeaters, means for equalizing transmission and optimizing signal-to-noise performance, equipment for monitoring, measuring, and fault locating, and multiplex to assemble the independent channels into a broad-band multichannel signal. The block diagram of Fig. 5 summarizes the system description built up in the previous paragraphs.
When a new system is to be installed, cable is loaded from a cable factory into the cable tanks of a cable ship (see Fig. 6). At the time it is loaded, the two ends of each cable section are brought as bights out of the cable tank and over to an area of the cable deck where the repeaters and equalizers will be stacked. When these have been loaded on board and stacked in racks, the cable ends are spliced to the repeaters and equalizers into one continuous subsystem (see Figs. 7 and 8). Needless to say, great care must be exercised to do this splicing right to avoid creating the world's longest and most expensive knot.
When the entire shipload has been assembled, it is powered and tested to verify that everything is as it should be. Automatic transmission measuring sets and sets for measuring the supervisory tones from the oscillators in each repeater are available, as is a small computer to aid in reducing the massive amounts of data which are generated in this process. Since the ships do not have the capacity to hold an ocean-spanning length, a system must be layer in segments. The end of the previous shipload was left buoyed off and, with luck, the ship will find the buoy still there when it returns (if not, the end must be grappled off the bottom). The new load is spliced on and laying commences. Under good conditions of weather with a flat ocean bottom, laying can proceed as fast as 8 knots. Since repeaters in modem systems are no longer flexible, they are not layed around a sheave, but through a linear cable engine which can accommodate a repeater when it comes along (see Fig. 9). After one repeater has been layed, the next one is moved from the stack into the launching chute.
As indicated previously, the cable is layed powered and is monitored continuously as it is layed. Transmission is between the shore and a transmission lead that comes out of the next equalizer to be layed. Interruption of transmission or even a sudden change alerts the laying team to possible trouble. Usually it turns out to be a measuring error, but every now and then it is a real problem. The process of laying has been aptly described as long periods of exquisitely boring routine interspersed by moments of sheer terror. As the cable moves from the ship's tanks to the sea bottom, its transmission changes gradually due to the effects of pressure, temperature, and handling. For the most part, these have been anticipated. The previously mentioned switchable networks in the ocean block equalizers are there to take care of any unanticipated changes. Measurements at regular intervals permit extrapolation of a predicted sea bottom transmission characteristic and, a few hours before an equalizer is due to go overboard, an optimum equalizer setting is selected. (Since there is no simple criterion for "best" setting, the process described by that short phrase sometimes involves heated debate and participants have been known to go to bed in a huff, mumbling extremely low opinions about their colleagues.) The selected setting is obtained by means of a stepping switch inside the ocean block equalizer which is activated by dc pulses into a stepping lead coming out of the equalizer. The transmission equipment is now connected to the transmission lead from the next equalizer. The transmission and stepping leads of the equalizer which have just been set are now molded over to seal them physically and electrically from. the wet high-pressure environment they are about to enter - and the equalizer is ready for laying. Second thoughts about the selected equalizer setting at this point are not welcome.
When the other side of the ocean is reached, the last cable aboard ship is connected to cable from the shore by a final splice. The system is now powered terminal-to-terminal and is lined up. This involves final equalization, signal-to-noise optimization by means of noise power rate tests, and acquisition of initial data required to administer the system. It is now ready for service.
See, lady in Philadelphia, I am not a drunk or a nut.
Robert L. Easton received the B.A. degree from the University of Chicago, Chicago, Ill., in 1950, and the B.S. and M.S. degrees from the California Institute of Technology, Pasadena, in 1953 and 1954, respectively.
He joined Bell Laboratories in 1954, and has spent most of the interval between then and now in the design and installation of successive generations of submarine cable systems. In particular, he has been involved in the area of system design, data processing, and computer aids for submarine cable systems. For a few years he taught the transmission System Design Course in the Bell Labs Communications) development Training Program, in the process editing the first hard-covered edition of Transmission Systems for Communications by Members of the Technical Staff.
Mr. Easton is a member of Tao Beta Pi.
Last revised: 24 December, 2009