This chapter is also available for
downloading. Please choose Word 97 format
or plain text and download the images for this chapter which are contained
in the file images1.zip.
|Introduction and Chapter 1
in Microsoft Word 97 format
|Introduction and Chapter 1
in plain text format
|Introduction and Chapter 1
images contained in zip file
format click here.
On each telephone call, a talking path must be set up between the calling and the called telephone. The method of making this connection, known as switching, has progressed from the simplest of hand operated switches through the more complex manual systems to the present mechanical switching systems.
This Survey of Telephone Switching describes broadly the characteristics of the major switching systems used by the Pacific Company. It is intended to give an appreciation of switching systems rather than details for engineering. The Bell System Practices are the "working manuals" used by engineers and provide the standard to which the switching systems are constructed.
No originality is claimed, as the material has been compiled principally from Bell System publications, and training courses of other Associated Companies.
DEVELOPMENT OF TELEPHONE SWITCHING SYSTEMS
In the development of dial telephone switching two fundamentally different arrangements have been devised for controlling the operations of the switches. In one, the switch at each successive stage is directly responsive to the digit that is being dialed. Systems using this method of operation are called direct dial control. An example is the step-by-step system in common use today. In the second arrangement, the dialed information is stored for a short period of time by centralized control equipment before being used to control the switching operations. The systems using the latter arrangement are known as common control systems. Examples of these are panel and crossbar.
A history of the evolution of these types of switching systems is presented, together with a discussion of their comparative merits for various fields of use and a look toward the future developments to be expected.
The need for a switching system became apparent as soon as public interest in telephony was aroused. This occurred in 1877, less than a year after Bell's original patent had been granted. The obvious method was to connect every telephone individually to every other telephone. But this procedure would quickly become complicated and impractical. Over a thousand different connections would be required to interconnect only 50 telephones.
The logical way to overcome the problem of interconnecting telephones was simply to gather up their connecting wires and bring them together in a central place. Then all that would be required would be a means of connecting the right telephone wires together when people wanted to talk. Upon this ability, the ability to interconnect any two of a great number of telephones, rests the value of telephone service to the public.
The first commercial telephone switchboard was placed in service in New Haven, Connecticut, in 1878 with a capacity of eight telephone lines. Within a year, the Connolly and McTighe "girl-less" telephone system was patented. The first patent on the Strowger step-by-step system was issued in 1891. By 1892, the first commercial installation of automatic switching equipment using step-by-step mechanism was made at La Porte, Indiana.
Though the rotary dial was developed in 1896, many of the early systems did not use this type. Various calling devices were used for a number of years. Among these were lever operated pre-set devices, key sets of several types, and dials with holes (in one case as many as 100) in which a peg could be inserted to act as a stop for an arm which was pulled around and allowed to restore. In all the early systems, regardless of the device used, the signals generated at the calling station directly controlled the selectors.
By 1905 there were several single office cities which had commercial installations of Strowger step-by-step equipment. A number of Western Electric Company 100-line and 20-line automatic systems were in service and a small amount of semiautomatic equipment under direct control of the "A" operator's dial was in operation. Planning was under way to remove some of the limitations and extend the field of use of the automatic and semi-automatic systems.
The switch mechanism employed in the step-by-step direct dial control system is arranged to take I to 10 steps vertically and I to 10 steps horizontally, thereby permitting the switch to have access to any one of 100 positions. This is accomplished by means of a bank attached to the frame and a set of wipers mounted on a shaft. The bank consists of ten rows, or levels, of ten positions each and placed one above the other to compose a 100 position unit. The shaft which mounts the wipers can be lifted and rotated so that by first raising it to a particular level and then rotating it,, the wipers can be placed in contact with any desired position in the bank.
The train of selection in a step-by-step office (Figure 1-1) is built up as the digits of the desired number are dialed by the calling subscriber. It is essentially a progressively built-up connection. The switches are selected in sequence and held until disconnection of the call.
Evolution of Principle of Translation
While mechanisms and circuits were being developed for direct dial control switching, work of a theoretical nature was going on which was to have an important effect on future designs. This work consisted of traffic probability studies and observations resulting in the development of formulas and curves examining the efficiency of trunk groups. These studies strongly influenced the views of engineers as to the economical size of switches.
Figure 1-2 is a reproduction of three curves produced by E. C. Molina in 1908, showing the average load carried by various numbers of trunks for three probability conditions namely P.01, P.001, and P.0001, corresponding to all trunks busy conditions encountered by calls once in a hundred, once in a thousand, and once in ten thousand times respectively. From these curves it can be seen, for example, that ten trunks can carry a load averaging slightly over four calls with a probability of encountering a busy condition once in one hundred attempts. Twenty trunks can carry an average of over eleven simultaneous calls with the same P.01 loss due to all trunks being busy, but with an increase of efficiency for the larger group of 15 per cent. The larger trunk group provides an increase in efficiency from 41 to 56 per cent.
These studies had considerable effect on the trend of overall system design. For example, it appeared that grouping subscriber lines on the connectors in groups of more than 100 might result in some economy. Other economies were possible if the limitations imposed by decimal selection could be avoided.
Direct dial control equipment is limited by the decimal digits dialed into it and by the speed of the customer's dialing. At each switching stage two actions take place. First, the switch follows,the dial pulses until it reaches a group of outlets corresponding to the dialed digit. Then, in the interval following this digit, and before the pulses of the next digit arrive, the switch hunts over the outlets for an idle path to reach the next stage. The number of paths from a switch level is, therefore, limited by the number of terminals the switch can hunt over in the interdigital interval. Direct dial control systems have generally employed switches with ten outlets per level. Special arrangements, such as twin levels, also have been employed to increase the number of outlets. A twin level switch provides terminals for two trunks at each rotary step and thus twenty trunks per level can be reached.
A new method of operation was necessary to take advantage of the Molina theory of trunk grouping and to offset the disadvantages of direct dial control. To allow time for trunk hunting over larger groups, and permit time for connections to distant offices, emphasis was placed upon the development of a system that could perform mechanically the functions of a manual operator. These involved answering the calling subscriber's line, recording and storing the called number, determining the route to the called party, selecting a trunk, and completing the connection to the called line.
However, another development,namely translation, was required before systems could operate with large access switches and non-decimal selections. Translation is the mechanical conversion of decimal information received from the dial to non-decimal forms for switch control and other purposes. When translation is made changeable, by some means such as cross-connection, it is the basis of much of the flexibility of common control systems.
A necessary feature of systems employing
translation of a series of digits, such as an office code, is digit storage. It was only a
small step from the concept of translation and digit storage to devices which provided
these features in common circuits. This development culminated in the design of the common
control system utilizing a "sender".
The Panel System
The panel system (Figure 1-3) is a common control switching system. It was developed for large metropolitan areas where the number of central offices to be served creates a complicated trunking problem. In the panel system, direct control of the switches by the subscriber's dial is abandoned in favor of a register, or sender, in which the pulses are stored until the equipment is ready to use them. This allows the selecting apparatus more time to hunt over large trunk groups than is normally present between the digits dialed by a subscriber.
The principal piece of apparatus which gives the system its name is the panel type selector, so-called because the terminals over which the selector passes are arranged in a flat rectangular bank or panel. This is used throughout the system in various forms, differing in size, in detailed arrangement, and in electrical connections, but all having the same general appearance and electromechanical construction. Each bank, employing punched metallic strips, can accommodate 100 outlets with three wires per outlet. Five banks are stacked into a frame over which 60 power-driven selectors can hunt.
The panel system uses revertive pulsing to control the selectors. With revertive pulsing, as the selector progresses, it sends back pulses which the sender counts. When the selector reaches the desired position, the sender stops it by opening the pulsing circuit. Because of the increased size of the switches, the panel system uses a continuously operated power drive common to a number of switches.
Development of a Large City Numbering Plan
By 1916 the full automatic system (Strowger) had established a competitive position with manual for single-office cities. Because the number of dial pulls for a single office was four or less, little concern was felt about dialing accuracy.
For the multi-office cities it appeared that full mechanical operation would improve service, be more economical, and reduce the pressing need for operators. However, in spite of these factors urging the adoption of a dial system, and even though automatic equipment was actually used in Los Angeles and Chicago in the first decade of the century, there was a reluctance to adopt full automatic operation in the very large multi-office cities because of the lack of a suitable numbering plan. An awkward plan was under consideration for handling dial traffic in these cities. It required the use of seven digit numbers with the dial customers being called upon to use arbitrary three digit numerical codes for the office names. At the same time, the existing office names would be retained for use by the manual customers. Adoption of this dual arrangement would have required the provision of a cumbersome directory. But worse than that, it was felt that dialing seven numerical digits would be too confusing to customers and consequently there would be an excessive number of dialing errors. It was, therefore, planned to use semi-mechanical operation for large cities, retaining an operator between the customers and the machine. While this scheme did not save as many operators as the full mechanical method, it was believed necessary to have trained operators so that the customers would not be subjected to the complications of dialing.
However, in 1917 W. G. Blauvelt of the American Telephone and Telegraph Company proposed a numbering plan which would permit the customers to dial up to seven digits with acceptable accuracy and which would also be satisfactory for manual operation. This arrangement consisted of the use of one to three letters and four decimal digits. The first one, two, or three letters of the office name were printed in bold type in the directory as an indication to dial customers that these were to be dialed ahead of the four digits. Manual customers used the office name as before. Letters as well as numbers were placed on the dial plate in line with the finger holes of the dial. This proposal was immediately adopted and further Bell System development proceeded along the lines of full automatic operation.
The Bell System planned to use panel equipment in large cities, not -only because of the trunk efficiency which was possible with the use of the large panel switch, but also because trunking, being no longer under direct control of the dial in this system, was divorced from numbering. The panel system was also attractive because it had flexibility for growth and for contingencies such as the introduction of new types of service. These advantages would be provided by the common sender and translators of that system.
Evolution of the Marker Principle
In retrospect, it is obvious that the developmental thinking up to the early 1920's was limited by the belief that it was necessary to have the selectors do the testing for idle trunks even with common controls. This arrangement had been successfully used in the step-by-step system and it was natural to follow the same plan in the panel. Subsequent development of the common control idea, started with an experimental "coordinate" system in 1924. Though the coordinate system was not developed for commercial use, it provided the marker systems in which the trunk testing is done by the markers.
The coordinate system derived its name from the method of operation of its switch, the process resembling the method of marking a point by the use of two coordinates. The switch was essentially a large version of the present crossbar switch and selected and held a set of crosspoints by the operation of horizontal and vertical members. Translation of the called office code, selection of a trunk, and operation of the switches to connect a transmission circuit to the trunk were functions of a new circuit, the marker, which the sender called into use for a fraction of a second after it had received the office code digits.
When the marker does the testing for idle
trunks, the trunk access from a particular switch is no longer a limiting factor in the
size of the trunk group. Thus it became possible to design systems using markers to do the
trunk testing and any type of switch to do the connecting. When a trunk has been selected
by the marker, the appropriate switches can be operated to connect to the marked terminal.
The maximum size of the trunk group need no longer be limited by the number of terminals
on one switch. With a primary-secondary switch array, groups much larger than those
accessible on a single switch can be handled.
In the early 1930's the Bell System started development of the No. 1 Crossbar system, with markers in both originating and terminating equipments and with improved features over the coordinate system which it resembled in many respects. A typical common control arrangement for a system using translation is shown in Figure 1-4 for No. 1 Crossbar.
Self-checking circuits, second trials and trouble indicators, which had proven highly successful in the later (decoder) variety of panel system, were important features of No. 1 Crossbar. Automatic alternate routing and the ability to operate with non-consecutive PBX assignments were major new features introduced for the first time in this system.
The subsequently developed No. 5 Crossbar system included a number of improvements, the chief of which from a common control standpoint was the use of common markers for originating and terminating business and the use of the call back feature in setting up the connection. In this system the common equipment records the calling line identification as well as the called number, and after setting up the path to the called line or outgoing trunk, breaks down the connection to the common equipment from the calling line and then re-establishes a connection back to the calling line.
Common control systems have been employed by the Bell System in addition to those already mentioned. These include panel sender tandem, crossbar tandem, and No. 4, A4A and 4A toll crossbar.
Even with all these improvements added the dial switching system, as it is known today, is an efficient machine only to a certain point. Registers, senders and markers can be furnished only in quantities that are reasonable and economical, based on normal busy-hour loads. When presented with an abnormal load, the system may fail because of faulty customer usage, such as failing to await dial tone.
The remedy for this, obviously, is a system so rapid in operation that registering devices are always available. Switching systems that perform at the speed required are not uncommon in other fields.
Other Digital Systems
In any discussion of telephone switching it is not possible to ignore high speed switching systems used outside the telephone industry. Other digital systems, which like a telephone switching system, receive and store information and manipulate this information toward a useful end in accordance with the rules of logic built into them have already acquired impressive proportions. Well known examples are the large scale digital computing machines that furnish answers to mathematical problems previously beyond the resources of human calculators. Also in operation are railroad and airline seat reservation systems and other similar automatic inventory systems that keep accurate track of thousands of items and furnish instantaneous answers to complicated inventory questions. Studies are being made of automatic air traffic control systems and there are in limited use automatic production lines, machines for weather prediction, for library search, for language translate on.
The new art of mechanized computing has already surpassed the switching art in one important way. Most modern large computers rely heavily on electronics, but modern switching offices do not. There are good practical reasons for the gap. Reasons of history, economics, and reliability constrain telephone engineers to a conservative approach. Any new system must be compatible with every old system. In other words, it must understand signals sent by the old systems and send signals understandable by them. In addition, the system has to work with existing wires and telephone sets in the presence of trouble and misuse. Also, newer techniques must demonstrate economic as well as technical superiority. Finally, systems must continue to operate reliably despite some equipment failures. Even the most optimistic electronic computer designers cannot yet claim 99 per cent, let alone 99.9 per cent reliability for continuous operation.
Despite these considerable difficulties, it seems probable that the next big advance in practical switching will be electronic. Electronics has high speed to offer the telephone engineer. This speed can be well used in a large exchange to handle the flood of control operations necessary to keep up with telephone traffic. As many as a dozen relay markers have been used in a No. 5 office. One sufficiently reliable electronic marker could replace all of these. Of course, such a simple replacement may not give the best design. It may be better to redesign from the ground up.
The Advantage of Speed in Switching
A picture of how speed could be used to advantage in a large exchange may be had by thinking first of a manual exchange which is small enough so that a single operator can handle all calls. Responding to each calling subscriber's signal she plugs into his jack, listens as he gives the desired number, then sets up the other end of the same cord to this number. She must also take down both ends of the cord when the circuit is no longer in use, handle calls to other exchanges on a special basis, provide information, record charges, and perform a variety of other duties. If the number of subscribers on this small exchange is increased, they would normally be served by adding more operators. Imagine, however, an operator who is human in every way except that she can operate at electronic speeds. She would not be able to take care of a much larger number of subscribers, since she would not have time to listen patiently as each calling subscriber repeated the number he wished. At ten calls a second, common in a large office during a busy period, if it required five seconds on the average to give each number by voice, at least fifty operators would be required even though all other functions could be compressed into zero time.
This fast electronic operator could then, obviously do little without something more to help her. If she could have some kind of bulletin board, on which calling subscribers could write, through dialing, in their slow and ponderous human way the calls they wished to make, and if she could read this bulletin board in her fast electronic way, then sufficient speed would enable her to set up and take down each call and still have time to spare for other necessary tasks.
It is apparent in the foregoing that electronics can be used to build a single control unit fast enough to handle the entire exchange on a "one-at-a-time" basis. It is also evident that electronic storage will be a feature of this electronic exchange and that electronic speeds may accelerate the switches themselves.
Electronic devices have advantages for switching other than speed. One advantage is size. Some of the most valuable space in the world is occupied by floor above floor of switching equipment in the downtown sections of large cities. Electronics may ultimately cut this space to a fraction of its present size. Another advantage lies in lower power consumption. Most relays absorb about a watt of power. A semi-conductor information-processing circuit containing a transistor and a few germanium diodes absorbs only a few milliwatts and can handle far more information per second.
A third advantage is reliability. Despite the high level of reliability and low cost which relays have achieved, they are essentially complex devices since they rely on a combination of electrical, magnetic, and mechanical phenomena. A relay which can outlast a billion operations is an engineering marvel. A megacycle electronic counter performs a billion operations in just 16 minutes and 40 seconds. Preliminary results indicate that solid state electronic devices can be designed to last many years. The expectation of failure per operation is obviously far less with these devices than with relays: a fact which, when combined with proper system design, should substantially reduce maintenance troubles.
Within the past several years, rapid progress has been made in the development of electronic devices that are adaptable to the rigid requirements of telephone switching functions. A development program is now active in the Bell Telephone Laboratories to apply them to the creation of a new switching system, the Electronic Central Office (ECO), that will make little or no use of electromechanical relays and switches employed earlier. The new switching system promises substantial economy in equipment costs and as much as 80 per cent saving in floor space. Economy is also expected in outside plant due to the use of two stages of concentration outside of the central office and to the higher loop resistances permissible. Operational speeds will be increased, maintenance will be made easier and the usual administrative activities will be simplified. The Electronic Central Office will more readily permit the introduction of new features and provide for greater flexibility in meeting changing requirements.
It is planned to make a trial field installation of an Electronic Office about 1958 for service of some 2000 lines (4000 main stations). By selecting a small relatively isolated single office town for the initial installation it is possible to restrict the area of intensive development for the present. Continuing development, however, will ultimately make the Electronic Central Office applicable to all switching requirements both with respect to size and to the services to be provided.