Reference / Paper · 1959
Bell Laboratories Record, January 1959 (Crossbar Circuitry for a Small PBX)
The 756A's marker, register, junctors and link wiring.
Reference / Paper · 1959
The 756A's marker, register, junctors and link wiring.
January 19E9 Bell Laboratories RECORD Development of the DEW Line Semiccnduct.iirs in Strain Gauges Flexible Repeaters for Transatlantic Cable Analog Comçuter for Evaluating Radar Broad-Band Dscilloscope -An Application Crossl ar Circuitry for a Small PBX Editorial Staff Editorial Board G. E. Schindler, Jr., Editor W. W. Mines, Associate Editor A. G. Tressler, Assistant Editor, Murray Hill J. J. Gruesen, Assistant Editor R. F. Dear, Design and Production T. N. Pope, Circulation Manager F. J. Singer, Chairman J. A. Burton J. W. Fitzwilliam F. A. Korn E. T. Mottram R. J. Nossaman W. E. Reichle THE BELL LABORATORIES RECORD is published monthly by Bell Telephone Laboratories, Incorporated, 463 West Street, New York 14, N. Y., J. B. FisK, President; W C. TOOLE, Secretary; and T. J. MONTIGEL, Treasurer. Subscription: $2.00 per year; Foreign, $2.60 per year. Checks should be made payable to Bell Laboratories Record and addressed to the Circulation Manager. Printed in U. S. A. © Bell Telephone Laboratories, Incorporated, 1959. Bell Laboratories RECORD Volume 37 Number 1 January 1959 Contents PAGE 2 Development of the DEW Line 7 Semiconductors in Strain Gauges io Flexible Repeaters for the Transatlantic Telephone Cable G. R. Frantz W. P. Mason W. M. Bishop and W. A. Klute 16 An Analog Computer for Evaluating Radar Performance C. C. Willhite and J. W. McIntyre 20 Broad -Band Oscilloscope -An Application 23 Crossbar Circuitry for a Small PBX R. D. Williams 28 J. B. Fisk Elected President of Bell Laboratories; M. J. Kelly, Chairman of Board; E. I. Green, Executive Vice President Cover A. F. Dietrich Inside a DEW -Line "radome," showing part of search antenna and protective geodesic cover. Story on DEW Line begins overleaf. Foremost among the contributors to the technology that evolved the Distant Early Warning Line are the Western Electric Company and Bell Telephone Laboratories. These organizations have been chiefly responsible for the development and installation that has led to the effective detection and communication systems for the DEW Line. G. R. Frantz Development of the DEW Line The DEW Line is a radar warning system that extends from the eastern coast of Canada to the northwest tip of Alaska. An additional section is now under construction along the northern coast of the Alaskan peninsula, extending to the Aleutian Islands. These two sections are connected by the Alaskan Aircraft Control and group of radar -warning Warning System stations which provide air defense for Alaska itself and early warning for the United States and Canada. The accompanying map (page 4) shows the position of the DEW Line and the other significant warning networks, the Mid -Canada Line and the Pine Tree Line. The objective of the warning system is, of course, to detect winged aircraft reliably, whether such aircraft are flying a few feet above the surface or at high altitudes. Having detected such targets, the system must then report the tracks accurately and promptly to air -defense headquarters in this country and in Canada. The Alaska- Canada section of the DEW Line was placed in operation on schedule (July, 1957) and is now reporting aircraft penetrations. The sea flanks are covered by airborne and U. S. Navy picket -ship radars. Future developments in technology will result in continuous improvement in this radar net. -a The Warning Line was conceived by the Summer Study group -which in 1952 met at the Lincoln Laboratory, Massachusetts Institute of Technology, to consider the problems of the evolution and further development of the air defense of North America. This group of scientists was composed of representatives from various government agencies, military commands and a number of private organizations, including Bell Telephone Laboratories. As the first step in carrying out the recommendations of this group, Western Electric was asked in December, 1952, to undertake the construction of a trial segment of the Line along the northern coast of Alaska. This experimental line was to be used for a field trial of detection and communication systems under Arctic conditions. Operating toward a critical deadline, Western Electric arranged for procurement of supplies, design of buildings and air strips, transportation of equipment, and actual construction of the stations. Bell Laboratories was asked to plan the electronics system and to arrange a series of important tests tests designed to demonstrate the technical feasibility of a detection line in the far north. Laboratories groups therefore procured special communications equipment, - 2 Bell Laboratories Record A typical DEW Line site. On either side of the spherical radome housing the automatic search raclai are the two sixty -foot antennas. At left can be seen the pair of thirty-foot parabolic antennas. equipped available search radars with automatic alerting equipment, and planned extensive tests and test instructions. Special "fence" type equipment was also procured automatic units for detecting aircraft in the areas that would otherwise be gaps between larger stations. With the assistance of Lincoln Laboratory and other organizations, these jobs were ready on schedule. By July, 1953, the equipment necessary for seven stations was loaded at Seattle for the voyage to the north shore of Alaska. The experimental stations consisted of a main station, two auxiliary stations, and four intermediate stations. The main station had detection equipment, lateral (east -west) communications, and rearward communications to a base station near Anchorage, about 650 miles from the Line. The auxiliary stations, spaced about 90 miles on each side of the main station, had detection equipment and lateral communications. The intermediate stations, which did not require continuous attendance by operating personnel, used the "fence" equipment to fill in the gaps between main and auxiliary stations. As a result, the over -all system had stations about every 30 miles. Western Electric completed installation of electronic equipment during the fall and winter of 1953, and tests began early in 1954. As expected, some of the equipment performed according to plan, and some was found to require improvement. For example, the Laboratories specified a redesign for the search radar to improve both its accessibility and performance. One result was automatic equipment for monitoring the performance of the system. This monitor sounds an alarm when the power falls below a particular value or when noise becomes excessive. For the other detection system the Laboratories specified improved performance under conditions of vibration, small fluctuations in power supply, variable propagation and moving clouds. Fur- - January 1959 3 thermore, to improve the performance of the person operating the console, suggestions were made to supply him with equipment for transmitting messages automatically. In general, the tests successfully demonstrated that a simplified, largely automatic system could be operated by a limited number of well- trained technicians, despite rigorous climatic conditions and difficult supply routes. The Air Force concluded, therefore, that a DEW Line extending completely across the Arctic was practicable, and late in 1954 asked Western Electric to build the complete line. As prime contractor, Western Electric recruited specialists from the entire Bell System and made arrangements with construction companies and with hundreds of suppliers and subcontractors. An important part of the project was selecting the route tremendous undertaking. Individual sites were chosen and the stations were laid out using criteria specified by the Laboratories. These criteria, besides the usual requirements for radars and communications, included many others imposed by the nature of operations in the far north : for example the need for fresh water, access to the sea, and surfaces for air strips. In designing the detection and communication system, Laboratories engineers drew upon recent -a iVV 411V 14 SOCIAL SOVY'ET gRCTICIRCL I r f 60 70° r70° ; , 11 , r rr_Mlo:i'iAiiWíV I ///v° / W,.a.__. _ Di - 6 yf T A LINE 4 advances in the art and upon their experience with the experimental network. The basic plan called for automatic search radars at auxiliary stations about 100 miles apart, with intermediate stations as "gap- fillers" employing another type of detection system. The detection information is passed laterally along this line, via UHF "beyond- the-horizon" circuits, to the main stations which have all the facilities of auxiliary stations. They also provide communications via VHF transmission to the mid -Canada line. In addition to these basic systems for detection and communication, each station has fixed station, mobile -radio equipment for communication with vehicles and with intermediate sites ground-to -air communication equipment in HF, VHF and UHF bands ; low frequency beacon ; and HF radio equipment for emergency use. Some of the electronic items are standard commercial or military units purchased by Western or supplied by the Government. Other equipment, however, required special development for example the detection units, lateral communications equipment, the control console for a station, and certain special test gear. For these, Bell Laboratories monitored the design and production or, in the case of the control console, actually designed the equipment and prepared for its manufacture by Western Electric. The control console presented a particularly difficult design problem, since it is the interconnection point of all the diverse communication and detection systems. Because of the complexity of the system, however, great emphasis was placed on ease of operation. Consider, for example, how the console operator at an auxiliary station uses a unit termed a "message composer." When the operator hears an alarm telling him that an aircraft is penetrating the network, he looks at his display equipment. From this he gets the aircraft's course, altitude and speed. He then places this information in the message composer by setting appropriate dials. It is transmitted along the lateral communication system to the main station to be plotted along with data from other auxiliary stations. At the main station, the plotting board enables the operators to filter out the "friendly" tracks, to correlate data from the various radar stations, and to determine on which tracks information should be transmitted to higher headquarters to the south. From these headquarters, : - LEFT: the three lines are positioned for complete radar coverage and dependable communications. January 1959 5 DEW Line terrain son' t i ntes presents difficult or unusual transmission problems. Here, the lateral system uses higher- powered, sixty-foot antennas. the main stations receive such important information as identification codes and flight plans. Equipment developed by other organizations was to operate reliably for long periods under Arctic conditions. Antennas and other exposed equipment were designed for temperatures ranging from -65 to +65 degrees F, and for wind velocities up to 150 miles per hour. The exposed equipment is able to operate under conditions of snow, rain or salt-laden air, and it can withstand heavy coatings of ice or snow. The indoor equipment, however, is not subjected to conditions significantly more rigorous than those normally expected for indoor, ground -based electronic systems. Indoor designs emphasized automatic operation, trouble -free performance, accessibility of components for maintenance, and ease of operation. The automatic search radar is a dual system with two antennas operated "back -to-back," each normally supplementing the other. During maintenance or emergency periods, either antenna may be arranged mechanically to provide coverage while the other half of the system is out of operation. This antenna, shown in the photograph at the The automatic search radar, shown here before the radome has been erected, uses two antennas of a special design, operated back -to -back. top of this page, is a scaled -up version of an earlier Bell Laboratories design used on another system. The radar proper is an extensive redesign of the system used on the experimental line, which was a good system but one designed for portability and therefore not ideally suited for permanent installation. The radome enclosing the antennas is built of Fiberglas and plastic sections that are specially shaped in accordance with a "geodesics" design. This rigid type of radome was chosen as a result of tests made by Lincoln Laboratory on the summit of Mount Washington. Full -sized radomes were tested in Labrador and Greenland. The "fence" detection system, as mentioned earlier, fills the gaps between radar stations. The various parameters of this system were developed after Laboratories engineers carefully analyzed the problems encountered on the experimental line. Particular emphasis was placed on reliability and freedom from false alarms in the automatic alerting features of this system. Special antennas were furnished to give direction of aircraft movement and to eliminate indications of slowly moving objects, such as birds. The lateral communication system operates "over- the-horizon" in the frequency range of 800 to 1000 mc. This system provides up to 24 voice channels along the line, some of which are used for teletypewriter service. In general, one kilowatt transmitters are used with two 30 -foot diameter parabolic antennas, but certain difficult links use 60 -foot antennas and 10- kilowatt transmitters. These antennas are shown on page 5 and at the right. The same types are used in White Alice -the Alaskan communication systems (RECORD, August, 1958). The VHF rearward communication system is of an ionospheric scatter type that uses a certain layer of the ionosphere to reflect radio waves, and it provides several teletypewriter channels in each direction. High -power transmitters are employed with two antennas having space diversity. That is, the antennas are attached to one receiver which uses the better of the two signals. They depend upon reflection from the ground to establish a lobe aimed at a region in the ionosphere midway between the transmitter and the receiver. The angle of this lobe depends on the length of the path, and the angle is obtained by installing the antenna at the proper height. During the first half of 1957, extensive evaluation tests of reliability of communication circuits, detection capability, and data -handling methods were made. No major flaws were found in the system concept or performance, and the Alaska- Canada portion of the Line is in full operation at the present time. Further tests are currently being conducted during operation by Bell Laboratories and the Western Electric Company, and also by Air Force agencies. In general, the performance has been successful to date with no more than the usual number of problems expected from any new system. The lateral communication system transmits one thousand watts from 30 -foot parabolic antennas. 6 Bell Laboratories Record liesearchers, concerned tritl the properties of semiconductors in studies of transistor action, have measured the piezoresistive effect such materials as silicon and germanium. Unexpectedly, this effect was so pronounced that it permitted the construction of very sensitive devices for measuring strain- devices that have nearly all the qualities of an ideal gauge. i W. P. Mason Semiconductors in Strain Gauges One of the most important and most interesting problems in mechanics is the conversion of mechanical motion into an electrical signal. In the broadest sense this is, of course, a fundamental feature of the telephone or of any other microphone -like device : mechanical motion of a diaphragm, ribbon, or some other vibrating material results in a fluctuating current. But a microphone is only one of a wide range of mechanical-electrical converters or transducers. Others are used in phonograph pickups, in crystal filters, in roughness indicators, and in many types of gauges for measuring such quantities as tension, compression, acceleration, pressure, shear forces and torque. In this article we are concerned particularly with this last category of measurement devices, since certain semiconductors hitherto used chiefly in the fabrication of transistors and diodes have been found to possess a number of remarkable properties when used in gauges. Before we discuss these semiconductors specifically, however, it is helpful to gain perspective by reviewing briefly some of the problems of this branch of mechanics. Of the many materials that can be used as electromechanical transducers, the two most important groups are the piezoelectric materials (mechanical motion produces a voltage) and the piezoresistive materials (mechanical motion produces a change in resistance). Magneto- strictive materials wherein a mechanical motion causes a change in a magnetic field, which then generates a voltage in a surrounding coil are - - - January 1959 7 - another possibility, but these have certain disadvantages that limit their use in many applications. The voltage generated in a magneto- strictive device is proportional to the time rate of change of the magnetic field thus the generated voltage is proportional to the velocity of displacement rather than to the actual displacement. Furthermore, magnetostrictive materials are subject to eddy- current losses at moderate and high frequencies, which alter the relationship between velocity and voltage. Hence such materials are not advantageous for measuring force, torque or acceleration. With any of these types of materials, the ideal would be to have a device that is very sensitive to small forces, responsive both to static forces and to periodic forces over a wide band of frequencies, operative over a wide range of temperatures ; SHUNT IDGE CIRCUIT RESISTANCE 2 BARS OF GER CLOC/f SPi9/NG DISPLACEMENT Displacement gauge constructed with crystals of germanium. Such gauges are up to two orders of magnitude more sensitive than common types. The torsional transducer: torque is applied to the arm on the free end of a germanium crystal; the other end is firmly cemented to a backing plate. and humidities, and stable with time. It is also advantageous if the device is easily fabricated and is small enough to add only a negligible weight to that of the object being measured. In general, piezoelectrics like quartz and ADP (ammonium dihydrogen phosphate) perform well at high frequencies but are inadequate for static or low- frequency forces because leakage current masks the response. In addition, they lose effectiveness at high temperatures and high humidities. Also among piezoelectrics, the newer ferroelectric ceramics like barium titanate (RECORD, August, 1949; September, 1955) have better temperature and humidity characteristics, and they have the advantage that they can be formed into irregular shapes. They tend, however, to change with time, and they may depolarize under high stress. (For an application of the barium titanate ceramics to problems of acceleration and shock, see RECORD, March, 1956.) There is much need, therefore, for more stable and more sensitive materials that operate over wider ranges of frequency, temperature and humidity; and semiconductors have most of these advantages. Many metals and alloys exhibit a change in resistance with strain, and a length of metal wire, usually the alloy constantan, is the sensing element in the most common types of strain gauges. In the simplest case, the gauge could be attached to a long rod that is to be measured, and for stress, weights could be added at the bottom of the rod. With increasing stress, current passed through the gauge would change as a result of the changing resistance of the wire. A bridge circuit is placed across the sensing wire, and a value of strain is read from a calibrated dial. The sensitivity of such a gauge is described by its gauge factor, given by the expression dR /RS where R is the resistance of the unstrained piezoresistive element and dR is the change of resistance with strain S. As an example, if the piezoresistive material in the unstrained state has a resistance of 50 ohms but increases in resistance by 0.01 ohm at a strain of 0.0001 unit per unit length, the gauge factor would be 0.01/(50 x 0.0001) or 2. Most commonly used metal-wire strain gauges have gauge factors between 2 and 4. Somewhat higher values are obtained with platinum wire, and values up to 20 are noted for nickel. By contrast, germanium is so much more sensitive that gauges using this material have factors up to 150, and silicon permits factors as high as 175. F. T. Geyling and J. J. Forst of the Laboratories have used germanium for a displacement gauge so sensitive that the factor limiting the sensitivity is the normal vibrations of a building. A shunt resistance had to be placed across the ger- 0.28 0.24 0.20 0.08 0.04 o 0 50 100 150 200 250 300 350 TORQUE IN GRAM -CENTIMETERS Configuration of the torsional transducer. Graph shows calibrations at different biasing currents. 8 Bell Laboratories Record manium to eliminate the effect of these vibrations, after which the device was still about fourteen times as sensitive as the usual wire -type gauge. With mounts that are resistant to shock and vibration, the full sensitivity (about eighty times that of a wire gauge) can be realized. Thus, germanium, silicon and other semiconductors permit strain gauges that are about two orders of magnitude more sensitive than older types. But this is merely one of their advantages. Equally important for many applications is the fact that semiconductors can measure static and low- frequency stresses, as well as higher -frequency stresses up to the resonant frequency of the material. In these new gauges, stress forces can be quite large, since semiconductors are generally strong materials. They can also be used to measure hydrostatic forces merely by immersing the gauge in the fluid. Further, since most semiconductors are structurally and chemically stable, their characteristics should not change appreciably with age or under conditions of high humidity. Operating temperatures with germanium range to 600 °K, and with silicon to 1000 °K. These limits are set by the "intrinsic" temperatures of the elements, above which they lose their piezoresistive characteristics. Above the intrinsic temperature, thermal agitation is the main cause of resistivity and is not affected by stress. Certain intermetallic semiconductors, like gallium arsenide, with an intrinsic temperature of 1400 °K, should be even better in this respect. These semiconductor crystals have another advantage not possessed by wire -type gauges namely that they respond to torsional and shearing forces, and thus can be used directly to measure torque. R. N. Thurston of the Laboratories has constructed the torsional transducer shown in the drawing and photograph on the opposite page. This is a cylindrical crystal of germanium cemented to a solid support at one end, with electrodes numbered 1 to 6 attached as shown in the drawing. With a proper biasing voltage, current is caused to flow from electrode 3 to electrode 5 and from electrode 6 to electrode 4. Electrodes 1 and 2, attached midway along the cylinder, are normally at neutral positions, but when a torque is applied to the free end of the cylinder, a voltage is generated across them. This voltage is directly proportional to the torque, and for this reason, no balancing bridge is required. Besides its sensitivity, this transducer has the advantages that it responds to both steady -state and alternating torques, and that the angular displacement is negligible. - January 1959 9 J. .1. l rst aclin.</ing an «ttrn-scnsitire sc>wicnncl nctm hrain gan¡/r constructed with germanium. The device uses a cylinder of germanium 0.130 inch in diameter and 0.50 inch long. The graph with the drawing of this un_t is a calibration in terms of voltage generated across electrodes 1 and 2 with various amounts of torque, for three different values of biasing current. In experimental work at Bell Laboratories, crystals were prepared and cut so that their unstrained resistance was 55 ohms the same resistance as that of a common wire -type element. Thus, the same bridge circuit and other associated equipment could be used merely by substituting the new gauges. These advantages mean that semiconductors have nearly all the qualities needed for an ideal gauge. There are still extreme conditions for which no simple measurement device is adequate, but the new gauges offer such greatly expanded characteristics that research, development and testing groups are sure to find many uses for them. There is another interestïng aspect to these gauges, namely, that they are a good example of how pure research often results in useful, and sometimes unexpected, by- products. Extensive physical and chemical research has been conducted on the properties of semiconductors, which was the reason the piezoresistance effect was measured, and this research has resulted in a useful device in an entirely different field. - One of the most important steps in the development of the transatlantic telephone cable system was the design of a repeater that could be laid under 2.5 miles of ocean water by the use of conventional equipment aboard cable- laying ships. Such a repeater is costly and difficult to repair and must be designed and constructed with care for long life. W. M. Bishop and W. A. Klute Flexible Repeaters for the Transatlantic Telephone Cable Development and engineering on flexible repeaters for the transatlantic cable system, installed in 1955 and 1956, was begun in late 1952. The schedule called for the completion of this work in the spring of 1954 to provide sufficient time for the manufacture of the required repeaters. This tight schedule was practicable because of the well documented background of engineering information available from earlier work on repeaters suitable for undersea use. Previous work on repeaters of this type culminated in the installation of the 1950 Key West Havana cable system (RECORD, March, 1952). Although this system was only about 120 miles long, the engineering was done with transoceanic cables in mind. As a result, sound basic repeater circuit, component and equipment designs of proven reliability were available for the transatlantic project. As finally developed, this circuit has three amplification stages, and the cathode heaters for the three electron tubes are connected in series. The positive voltage drop across the three heaters is used as the plate supply for the electron tubes. As in all cable systems, the signal decreases in power with distance over the cable. Further, this attenuation of the signal occurs in a predictable pattern or "shape" of a curve plotted to show the amount of attenuation at various frequencies. Thus, the repeater must not only "boost" the over -all power, but also must compensate for the higher attenuation of the higher frequencies. This is referred to as "gain- shaping" to counteract the "cable-loss shape," so that finally, the signal is delivered with very nearly the original frequency -power distribution. The gain- shaping necessary to match the cable loss shape is obtained in approximately equal parts from the input and output networks and from the beta circuit (used to provide negative feedback to the amplifier). The crystal across the beta circuit affords a means of evaluating the performance of individual repeaters and of identifying inoperative repeaters during the life of the system. The Key West- Havana repeaters using this circuit handle twenty -four 4 -kc voice circuits between 12 kc and 108 kc. The plan for the transatlantic system specified that the repeater gain compensate for the loss of 36.9 nautical miles of cable in its ocean -bottom environment over as wide a frequency band as 10 Bell Laboratories Record was consistent with accepted performance requirements. Experience gained with the Key West- Havana repeaters under operation showed that repeater design margins could be reduced. This, together with the planned use of a larger diameter cable with lower loss, constituted the more important considerations leading to increasing the bandwidth of the repeater for the transatlantic system. The development of the electrical circuit to broaden the frequency band began with the basic Key West- Havana circuit. Initially, development effort was concentrated on the revision of the input and output coupling networks. These networks were scaled up in frequency from their original range of 12 to 120 kc to the range of 20 to 200 kc. The final design of each of these networks matched about one third of the attenuation of 36.9 nautical miles of cable. In addition, the output circuit was designed to present a load to the third tube of the repeater that was satisfactory from the standpoint of modulation performance. Also, the circuit permitted the tube to be operated at close to its maximum power capability at full drive. The interstage networks were then revised. These were also scaled up in frequency from the Key West- Havana designs. Also, the effective INTERSTAGE parasitic capacitance was reduced approximately 4 mmf. This reduction resulted in the lowest values obtainable from the processes used to manufacture components. The reward was an increase of approximately 2 db in the forward gain of the amplifier and about 1 db in the maximum available feedback. The beta circuit was the last circuit to be developed. This network, in addition to supplying one third of the required transmission shaping, was used to equalize all the small deviations of amplifier gain from cable loss. Any attempt to make each of these shaping networks exactly match one third of the attenuation slope over the frequency band would have introduced additional components in each network with associated layout problems. The beta circuit was therefore modified to compensate for deficiencies in the other shaping networks. Next, the simple beta circuit was modified to be applicable to the new frequency band, and additional elements were added to obtain the desired compensation for the cable. Residual deviations were sinusoidal in shape and hence were minimized by the addition of tuned circuits. A schematic of the final amplifier circuit for the transatlantic system is shown on this page. It is not markedly different from the one used for INTERSTAGE 2 I INPUT COUPLING NETWORK OUTPUT COUPLING NETWORK 1 000 BETA NETWORK I BETA NETWORK 2 r CRYSTAL Y The circuit used for the transatlantic repeaters. It compares with the Key West -Havana circuit except that the present design has a more elaborate January 1959 11 beta circuit plus a gas tube for protecting the power circuit. Repeater design consists of the amplifier circuit assembly and repeater container. extensive re- engineering of the repeater was required because it was to be manufactured on a production basis by the Western Electric Company. This involved the preparation of new manufacturing information in the proper form for factory use. The mechanical design of the repeater consists namely, the amplifier cirof two major parts cuit assembly and the container, including the remaining components of the repeater. The amplifier circuit assembly consists of 67 electrical components grouped in 15 rigid sections, plus a short terminal section at each end, which brings the number of sections to 17. Eight of the sections contain individual components such as power -filter capacitors or electron tubes. One section contains a desiccator. Eight sections contain the remaining electrical components. This makes a structure about 84 inches long with an outside diameter of 1.5 inches. The adjacent sections are attached to each other mechanically by axial helical steel springs to give the structure the desired mechanical flexibility. They are connected electrically by bus tapes which have slack loops between sections to provide flexibility. Each section includes one or more electrical and mechanical components enclosed in a plastic "container," which in turn is enclosed in a plastic cylinder. The bus tapes run in grooves between the plastic container and plastic cylinder. The latter unit must provide satisfactory insulation between the repeater circuit and the container for dc voltages up to 2500 volts. Besides the indicated mechanical and electrical design changes of components and the addition of two new amplifier sections, another important design modification was made. Previously, wherever metallic finishes were exposed to the repeater atmosphere, tin or zinc coatings were used. In the new design, these were changed to gold plate. This was done to reduce the amount of metal "whisker" growth (RECORD, November, 1954 and the associated danger of cross contacts within the repeater. As seen in the drawing on the next page, the repeater container consists of two layers of abutting steel rings each 3.i-inch long. The abutting planes of the rings in the separate layers are staggered a half ring- length to eliminate any gap from the outside to the inside of the container and to keep the rings in relative alignment. The steel rings are the medium that enables the container to resist ocean pressure. Placed over the rings is a thin-walled annealed copper tube. This tube, together with a system of end seals hermetically sealed to it, resists the - W. A. Klute (left) and W. M. Bishop examine an amplifier section. This view shows the unsheathed amplifier with adjacent sections attached to each. other; they are connected electrically by bus tapes. the Key West -Havana amplifier, but the beta networks have three more tuned circuits and some additional resistors. The repeaters employing this circuit provide thirty -six 4 -kc voice circuits between 20 and 164 kc. In addition, maintenance facilities for the system (order wire and printer circuits) are available below 20 kc. The fault- localization facility (crystal band) is provided above the message circuits in the range of 170 kc. A gas tube has also been added to the circuitry. Its purpose is to provide a current path in the event an electron -tube heater opens. The remainder of the repeaters will then continue operative for test purposes. The potential across the gas tube is approximately 50 volts when the heaters are normal and rises to well above the firing voltage of 150 volts in the event of an open heater. When the tube is fired, the voltage across it drops to about 10 volts. Over -all mechanical design of the repeater, as with the circuit design, closely followed that used in the Key West- Havana project. Modifications were necessary, however, partly because of changes in the electrical requirements of the circuit and components, and partly to take advantage of new developments in design. Furthermore, ) 12 Bell Laboratories Record entrance of ocean water into the repeater. In these respects, the repeater container for the transatlantic project is identical to that used in the Key West -Havana project. Other than the increase in length resulting from the additional amplifier sections, the most extensive change in container manufacturing information for the transatlantic project was the introduction of a new test for leaks. The previous method (RECORD, July, 1951) was considered impracticable for factory use. Consequently, a new two -step method of container leak testing was developed. In the first step, access to the inside of the repeater is retained via a single opening, and the exterior of the remainder of the product is subjected to helium at pressures comparable to those on the bottom of the Atlantic. The opening makes it possible to analyze the internal atmosphere for helium through the use of a special mass spectrometer sensitive only to helium. The presence of any leaks is thus detected and the leak rate is determined quickly. In the second step, after the access opening is closed, the product is again subjected to pressure for a known period. The pressurizing medium in this case is water and a solution containing a small amount of a gamma- emitting radioactive material. The solution surrounds the access -opening seal in such a way that, in case of a leak, only the solution will enter the product. The presence and the amount of the solution is determined, CORRO E. ARMOR BEDDING LAYERS AND TWO LAYERS OF ARMOR WIRES TYPICAL CONTAINER-TO-SEAL BRAZE after a pressurizing period of a few days, by radiation detectors which receive gamma radiation through the wall of the container. The new method has the advantage that the detectors detect only the leaking medium. The older method involved weighing the repeaters before and after pressurizing, and various factors (dirt or corrosion products and entrapped water) could contribute to errors in determining leak rate. The transatlantic repeater shown below was manufactured under the control of the Kearny Plant of the Western Electric Company in a special shop located at Hillside, New Jersey. There, most of the mechanical and electrical components were fabricated, and all repeater assembly operations except armoring were carried out. As in most projects of this nature, however, it was necessary to perform some operations elsewhere. The electron tubes were made and extensively tested by the Laboratories at Murray Hill, New Jersey. The glass seals were manufactured by the Western Electric Company at Allentown, Pa., with the close cooperation of the Laboratories people at that location. And finally, the repeaters themselves were armored by the Simplex Wire and Cable Company, located at Newington, New Hampshire. Although the repeaters for this project were manufactured by the Western Electric Company, the basic conditions, procedures and tooling followed closely those used by the Laboratories for the earlier project. This resulted largely from GLASS- TO - KOVAR COPPER CONTAINER TUBE SEAL CONNECTIONS TO AM'LIFIER r CORE I TUBE COVER SEAL CORE TUBE SEAL CAVITY PRESSURE LIMITER LIQUID INSULATOR (POLYBUTENE) I V STEEL RINGS GLASS SEAL IWa RUBBER SEAL Internal construction of an undersea telephone cable repeater. Repeater enclosure is terminated at each end with a system of seals: glass- metal, January 1959 13 a plastic seal, and lastly a 7 -foot long seal formed within a copper tube which is an extension of repeater housing. These resist sea -bottom pressures. The transatlantic cable shown here paying out off the stern of the cable -laying ship H.M.S. Monarch. The fifty-one repeaters had to be flexible enough to go over this wheel on their way dawn to the rugged Atlantic ocean bottom where they lie at intervals of 37 nautical miles. three considerations. The first and most important was the reliability requirement on the product. This required extreme care in workmanship and near clinical cleanliness during manufacture to minimize the danger of repeater failure due to contamination. Such a failure would result in costly repair operations. The second consideration was the size of the project. Since the required quantity of any specific component was in the tens or hundreds rather than in the millions, laboratory -like methods were practicable. Third, the unusual nature of the product, and the severe test re- quirements and urgency of the project, all made it desirable to adhere as closely as possible to methods which long investigation and experience had proved satisfactory. As might be expected in a project of such scope and importance, many people were involved. The successful operation of the transatlan