History of the Atlantic Cable & Undersea Communications
|The History of the Repeater
The History of the Repeater
Have you ever wondered why we call the submerged amplifiers that boost the signals on our trans-oceanic systems Repeaters? In terrestrial telecommunications, optical amplifiers are called In Line Amplifiers (ILA). The history of the repeater goes back to the infancy of the telephone era of the submarine cable industry but the word actually comes from submarine telegraphy.
Submarine telegraphy had yet to reach its zenith when, in 1891, the British General Post Office (GPO) laid the first subsea telephone cable of any note across the English Channel. This system was supplied by Siemens Brothers from its Woolwich factory. It used a telegraph cable design that limited transmission to relatively short distances, due to the distorting effects of the cable’s capacitance.
In 1896 subsea telephone cables were laid across the Solent and then the Irish Sea. These were manufactured by Telcon at Enderby Wharf in Greenwich. The problem of cable capacitance was overcome by a technique known as ‘loading’, which involved the inclusion in the cable construction of an alloy tape with special magnetic properties. Telcon supplied the first cross-channel telephone cable of this type in 1912. This same technique was also applied to telegraph cables. In 1924, Telcon supplied the huge US telegraph operator, the Western Union Telegraph Company, with a transatlantic cable between New York and Horta in the Azores, which was capable of a transmission rate of 1,500 words a minute. Further developments involving the use of Mumetal in the cable construction increased this capability to 3,000 words a minute by 1928.
In 1895 Telcon had been granted a patent for a subsea telegraph cable with a helically-wrapped copper outer conductor. However, this idea was not exploited until 1921, when Telcon made three coaxial cables and laid them between Havana, the capital of Cuba, and Key West, Florida.Just prior to the Second World War (WWII), submarine telegraph cables were struggling to compete with the faster, cheaper radio telegraph, and submarine telephony was still in its infancy. Thanks to the work of Oliver Heaviside (1850-1925) into skin effect, and his invention of the coaxial cable, plus the discovery of polyethylene by ICI in 1933, which overcame the high capacitance associated with gutta percha insulated cables, by the late 1930s it was possible to send telephone speech over approximately 100 nautical miles of submarine cable.
In 1947 the first commercial submarine coaxial cable was laid across the North Sea from the UK to the Netherlands. However, even with these lower attenuation cables, to achieve greater distances some form of amplification was required. The only available technology for amplification at that time was thermionic valves (known as vacuum tubes in the USA). The triode had been invented by the American Lee De Forest (1873-1961) in 1907 and the pentode by Dutchman Bernard D H Tellegen (1900-1990) in 1928.
These were the only components that were then available for the development of an amplifier which could operate in a water-tight housing for many years. The British were the first to attempt this when in 1943, as part of the war effort, the GPO used CS Iris to recover the existing Anglesey to Port Erin telephone cable, joint in an amplifier, and re-lay it. As a result of this experimental design, by 1948 coaxial systems containing a single submerged amplifier had been installed from the UK to Germany, the Netherlands and Denmark. However, these repeaters were cumbersome to handle and difficult to deploy because both ends of the cable were connected to the same end of the housing. In addition, the housings were only suitable for the relatively shallow waters of the North Sea. If telephone cables were to cross major oceans further development was required. On both sides of the Atlantic, this objective was pursued independently and resulted in two radically different solutions.
Early British Post Office Repeater
In parallel with submerged amplifier development, the ailing submarine telegraph industry was looking for ways to improve the performance of its cables. Western Union developed a submerged housing that contained thermionic valve circuitry to detect incoming telegraph signals and regenerate or ‘Repeat’ them, with newly generated outgoing signals. The first of these regenerators was successfully inserted into the 1881 American Telegraph cable in 1950. These devices were named ‘Submerged Repeaters’ and over the next decade several more were inserted into existing submarine telegraph systems. No new telegraph systems were ever installed that included Submerged Repeaters, but for some reason that name survived the demise of the technology!
One of the major technical problems that repeater designers had to resolve was how to integrate the repeater into the cable so that it could be laid by a cableship without major problems. The standard cable laying machinery of the mid-20th Century was the Drum Engine. This was a wide, 6 foot (1.83m) diameter drum around which 4-5 turns of cable were wound so that the surface friction of the drum provided sufficient grip for hold-back tension to be applied to the cable by the Drum Engine.
The US approach, led by Bell Labs, was to develop a long, thin flexible housing that could pass around the Drum Engine in the same way as the cable, without significant reduction in pay-out speed. The initial work on this design started as early as 1919 and was completed in 1941; however, because of WWII, the first production repeaters were not deployed until 1950, when they were used on the Havana to Key West system. These flexible repeaters were subsequently deployed on the first Trans-Atlantic Telephone cable, TAT-1, in 1956 and were also used on HAW-1 (California-Hawaii) in 1957 and TAT-2 in 1959. Because of the size and shape of the housing, these amplifiers could only be unidirectional, and so two cables were required to provide a complete circuit. This design achieved a system capacity of 36 x 4kHz voice channels.
British development, led by the General Post Office (GPO, now BT), resulted in an in-line 10½" (26.7cm) diameter rigid housing which had room inside for filters, allowing it to provide bi-directional transmission over a single cable. The British design could provide up to 60 x 4kHz voice circuits, nearly twice that of the US design, and being bi-directional it required only a single cable.
These housings were 9 feet (2.74m) long and were initially known as ‘Lump’ Repeaters, to differentiate them from the American flexible design. Because of the rigid housing, these repeaters would not pass around the Drum Engine. In order to deploy a rigid repeater the cable ship would have to be stopped, the cable taken off the drum, and the repeater fleeted past the Drum Engine using ropes and shackles. The cable turns would then be replaced onto the Drum Engine before pay-out of the cable and repeater could recommence. It was considered that in deep water this process would have a high risk of throwing cable loops on the sea bed, thus increasing the chances of cable faults. For TAT-1 this risk was considered too high.
The British continued to work on this problem, and in November 1956 the GPO approached Submarine Cables Ltd (SCL) with a proposal to jointly develop a cable engine that could lay rigid repeaters and the new lightweight (LW) cable which they were developing in partnership with SCL and Standard Telephones and Cables (STC), in a continuous operation without stopping the cable pay-out. The concept had been developed at the GPO Research Station at Dollis Hill, where scale and full-size models had been built to prove the design. The Post Office drew up a design specification and after some negotiation a contract was awarded to Telcon for the engineering, design, development, manufacture, land trials, and installation of the new cable engine on Her Majesty’s Telegraph Ship (HMTS) Monarch (4). The work had to be completed and the ship ready for sea trials in 14 months.
This new cable engine was designated ‘V-Sheave Gear’. It comprised six vertically-mounted 6 foot (1.83m) diameter wheels geared together; the inboard five wheels had a ‘V’ slot in the rim, while the outboard wheel had a ‘U’ slot. This sheave served as a guide wheel as well as a dynamometer to monitor cable tension and cable pay-out speed. The cable was fed around these wheels in a serpentine path and rested in the V slots. When external tension was applied to the cable, it was forced further down into the V slots, thus increasing the grip. The challenge, of course, was to be able to get the repeaters outboard of the engine without slowing the cableship too much.
Land-Based Trial Engine at the Telcon Factory in Greenwich
The solution was the use of a bypass rope which was attached to the cable in front and behind the repeater. The bypass rope passed through the V-sheave gear while the repeater was carried past on a trolley. This operation creates slack cable behind the repeater and in order to ensure that the outboard tension is not applied to the cable until it is fully re-engaged with the six sheaves, it is necessary to temporarily shorten the effective length of the bypass rope. This was achieved by what was called the ‘shortening’ or ‘compensating’ gear’. This comprised two vertically mounted sheaves set in the cable line such that a third sheave, driven by a hydraulic ram, could be lowered or raised between them to form an omega (Ω) shape in the bypass rope, allowing the repeater to be moved back into the cable line. The third sheave would then be disengaged once the cable was back in the cable engine and outboard tension re-applied to the cable.
HMTS Monarch (4) set out for Atlantic sea trials at the end of January 1958. These were successfully completed by 3 March 1958. Once these trials had proved that rigid repeaters could be safely be deployed in deep water, further trials were conducted on the new lightweight cable in preparation for the installation of CANTAT in 1960-61. Soon all British cableships were equipped with V-sheave gear, and this cable engine design remained the method of deploying rigid repeaters until the introduction of the Linear Cable Engine (LCE) in 1971. LCE was yet again a British development.
CS Alert V-Sheave Cable Engine, 1970
CANTAT-1 also presented another problem for the rigid housing design. As noted above, CANTAT-1 was the first system to use lightweight (LW) cable, which had the strength member on the inside of the cable structure. The coaxial cable design at that time was one inch (0.990”) in diameter and it was believed that having to support the weight of the repeater housing in the catenary to the sea bed would cause excess strain on the cable or cause cable run-away. A method was needed to relieve strain on the cable. The answer was to attach silk parachutes to the repeaters as they were deployed, the theory being that the parachute would open in the water column and bear some of the weight of the repeater during its descent to the sea bed. These parachutes were successfully trialled in Loch Fyne in 1960 and were used on all British repeaters deployed in deep water until the introduction of a new, stronger, one and a half inch (1.47”) cable design in 1968; after that the practice was abandoned.
SCL Repeater Housing
STC Repeater Valve Amplifier
STC Repeater Internal Units, 1954 Design
STC Repeater Housing
CANTAT-1 marked the end of the flexible housing design and from then on manufactures in France, Japan, UK and the USA all adopted the rigid housing. For the remainder of the Telephone Era (1950-1986) repeater mechanical design changed very little. Semiconductor transistors replaced thermionic valves in the early 1970s and system transmission capacity was improved through several iterations. System capacity was initially designated by the number of voice channels, and then by the highest frequency transmitted. The 80 circuits of CANTAT-1 soon became 160 circuits (1.2MHz), this was followed by system designs of 5MHz, 12MHz, 14MHz, 36MHz and finally 45MHz. The highest capacity submarine telephone system ever built was PENCAN 3, which was installed in 1977. It was designed to support 71 supergroups or 5,680 x 3kHz voice channels. With the advent of fibre optic transmission, the repeaters used in these early systems became known as ‘Analogue Repeaters’.
STC Analogue Repeater Being Laid
The telephone cable or analogue repeater used the following elements. The electronics was contained in a rigid cylindrical housing with water tight bulkheads at each end, capable of withstanding water depths of 8000m. At each end it had an anchorage system which could terminate a range of cable types. This anchorage system included rubber buffers that tapered to ease the passage of differing cable types through cable machinery. The rubber buffers provided bend limitation at the end of the housing and a taper to ease the passage of the repeater through the LCE. For at least one manufacture, these buffers supported sacrificial zinc anodes intended to protect the steel casing against galvanic corrosion. The cable passed through the bulkhead via a gland that prevented water penetration of the repeater in the event of cable damage close to it. Within the repeater, the DC power was separated from the AC transmission through power separation filters, which also contained surge protection circuitry. The transmission signals were separated into High and Low Bands using directional filters; the signals then passed through wide-band amplifiers. In early repeaters a single amplifier was used for both bands of transmission, but for the higher capacity systems there was a separate amplifier for each band.
The repeaters also contained a supervisory system; early versions allowed a unique signal to be sent from one terminal station, which could be detected by a single repeater, using very narrow band-pass crystal filters. This signal was amplitude modulated with an oscillator in the repeater and returned to the same terminal station in the other band. Later systems had crystal oscillators in each repeater which transmitted unique frequencies below the Low Band transmission spectrum and above the High Band spectrum. To illustrate this, if we consider the STC 14MHz system design, which was far and away the most popular (by sales volume) submarine telephone system, the frequency spectrum was as follows: Low Band Supervisory (200-300KHz); Low Band (310KHz-6.0MHz); High Band (8.0-13.7MHz); High Band Supervisory (13.7-14.0MHz).
As we moved into the Optical Era, transmission technology once again became unidirectional, but this time using single fibres within the same cable. The first generation of optical repeaters were based on regenerative technology; i.e. receiving a light pulse and then transmitting a new light pulse over the next cable span. This required the development of new components, such as lasers, photo-diodes and integrated circuits, with sufficient reliability to survive 25 years on the sea bed. Apart from the components, repeater designers faced five new challenges; first the anchorages had to be redesigned to accommodate the changes in design of the cable; secondly, the repeater bulkheads had to accommodate separate, watertight electrical and fibre paths; thirdly, there was no need for power separation filters but, due to the integrated circuits, surge protection became of primary importance; fourthly, as the attenuation of fibre is adversely affected by hydrogen ingress, care had to be taken in the choice of metals and coatings used on the repeater housings, in order to minimise local generation of hydrogen; finally the power consumption and heat generation within the housing increased significantly when compared with the previous analogue repeaters, and so heat dissipation was another major concern.
The capacity of optically regenerated systems developed over a decade and was characterised by the line rate. Initial system designs operated at a wavelength of 1,310nm (nanometers) and a line rate of 280Mbit/s and the line rate later increased to 420Mbit/s. The total system capacity was calculated as the line rate multiplied by the number of fibre pairs in the cable. By 1990 systems were operating at 1,550nm and a line rate of 560Mbit/s.
Alcatel Submarine Networks Optical Repeater
The latest iteration in optical repeater design occurred in 1995 when the first optically amplified systems, operating on a single wavelength at a line rate of 5Gbit/s, were deployed. The use of EDFAs as optical amplifiers in repeaters opened the door for WDM and DWM transmission and increased line rates without any changes being required to the design of the repeaters. It was roughly at this time that the capacity of submarine cable systems ceased to be calculated in terms of a number of 64Kbit/s voice channels and just became a total data capacity. The first generation systems used 1,410nm pump lasers, and then 980nm pumps were introduced for higher power and greater reliability. Initially four pump lasers per amplifier were used, but as confidence in laser reliability increased this was reduced to two pumps per amplifier. The submarine section of a modern optically amplified system is now close to transparent, allowing system capacities of many terabits; the capacity being defined by the equipment on the ends and being determined as follows:
(Number of Wavelengths) x (Line Rate) x (Fibre Pairs)
In many ways things have come full circle, because to achieve these capacities system design engineers must concern themselves with signal to noise ratio and wideband equalisation, just as their predecessors did during the Telephone Era when designing systems using analogue repeaters.
Article text copyright © 2016 Stewart Ash
Last revised: 26 October, 2016