Alexander Graham Bell (1847-1922) is well-known as the inventor of the telephone, and who demonstrated early instruments at the Centennial Exhibition in Philadelphia in 1876 on a 500-ft line. Bell clearly recognized the nature of speech signals and designed the magnetically biased receiver that became standard. The voice-powered transmitter was, however, quite feeble, permitting only short-distance communication, while the variable-resistance liquid transmitter used at the Exhibition was not suitable for general application. Thomas Alva Edison (1847-1931) developed the carbon-granule transmitter in 1883 (see Microphones for the history of microphones, and the origin of the name), which produced strong speech currents, making the telephone practical. Many other workers contributed to the development of the telephone, such as Gray, Berliner, Reis, Hughes, Hunnings and others. The arguments over priority and the patent battles are well-known and disgusting, but they did leave us lots of information. More kudos should adhere to the one who perfects an invention than to the one who merely conceives of it theoretically, but both deserve credit. Edison was a wonderful perfector.
The great advantage of the telephone is that it required no skill to operate, merely the capacity to speak and hear. There was no need to handle messages through a trained operator, and the service could be available wherever it was required, not only in special offices with inconvenient locations and opening hours and prices per word. This led to a great proliferation of lines connecting subscribers to a central office, where the instruments of those desiring to speak with one another could be interconnected. There were telegraph switchboards in a few places with large traffic, and these were used as a pattern for telephone switching, but normally the few lines were able to be connected as desired with something no more complicated than a plugboard. Telephones made switchboards absolutely essential. Even a score of subscribers made subscriber-to-subscriber lines a hopeless tangle. The switchboard was introduced in 1879.
At the present day, only the subscriber loop still retains most of the familiar features of the classic telephone. The switching is now all done automatically, and the signals are handled on 4-wire systems or their equivalent, with separate channels for the two directions, with the signals usually in digital form. The analog voice signals are digitized, and are reconstituted at the other end of the link. Telephone sets themselves are very different, using different transducers and electronic amplification. They are not much better than the classic telephones in quality, but are very much cheaper.
In this article, I will try to describe the classic telephone and switchboard working over full-duplex wire lines. My principal example will be the U.S. Army EE-8 field telephone, used in World War II, which is a descendant of the magneto telephones used for many years in the Bell system. Military communications then depended firstly on the written message and secondly on the telephone, the radio being used only when necessary. Radio communication can be established almost instantly over considerable distances, without the requirement of wires, but it requires considerable power, is subject to propagation disturbances, is not secure, and is easily jammed. With today's mobility, however, it is relied upon almost exclusively, and fails often to behave respectably.
The transmitter and receiver are the heart of a telephone set, so let's examine them first. In telephone work, the term "transmitter" is used instead of "microphone," perhaps the more common term. The transmitter is a transducer, producing an electric signal that more or less closely follows the pressure or velocity variations in a sound wave. These variations are remarkably small, so the transducer must be sensitive; that is, it must produce a reasonable electrical power output. Bell at first tried using the current induced when the magnetic field through a coil of wire was varied, since moving a wire in a magnetic field was far too insensitive. In fact, shouting into the Bell receiver, which will be described below, does generate an electric current. However, this was far too insensitive for communication over any practical distance, so Bell turned to transducers whose resistance varied with sound pressure. A needle dipping into an electrolyte and moved by a diaphragm proved adequate for demonstrations.
The search for a variable-resistance transmitter occupied many inventors, since the need was obvious, and a variety of devices were brought forward. Edison joined the search, and improved a transmitter devised by Hunnings, an English inventor who used granules of coke. Edison subjected it to his typical development process, and found that granules of anthracite coal were superior, among other things. He brought the transmitter to a certain state of perfection, and deserves full credit for this. Everyone looked at what everyone else was doing, and nobody could be called "original." This is the usual case with inventions in which the need is well-known. Patents can actually retard the progress of invention by excluding inventors with better ideas when a bad idea has been protected. Later, the Edison transmitter was improved upon in its turn.
A carbon
transmitter is shown in the diagram at the right. This schematically represents
the classic transmitter of the 1920's and 1930's as used in a desk set. The
principles are exactly the same in earlier and later examples. Carbon
transmitters used in later handsets were more compact, and the carbon chamber
was a different shape to facilitate use in different orientations.
The diaphragm is of thin aluminium, and presses on the front of the carbon button. A damping spring controls the motion of the diaphragm, and makes electrical contact. Insulation is provided by fibre washers and bushings, so that both the front and back of the carbon chamber are insulated from the frame and from each other. The carbon granules are from ground anthracite, passing a 60 per inch screen, but retained on an 80 per inch. They are carefully cleaned and heated in hydrogen to passivate the surfaces. The granules are held between carbon discs at front and back. This transmitter was sensitive between about 300 Hz and 3000 Hz, which determined the voice-signal bandwidth, and strongly peaked at about 1000 Hz. Later developments evened out the response considerably by 1927, with further smaller improvements continuing.
The carbon microphone is ideal for creating electrical noise, which appears as the "carbon hiss." The noise increases with the bias current, limiting the output of the transmitter. This noise is scarcely noticed in telephone applications, but is very disadvantageous in more critical ones, such as radio broadcasting and recording. In spite of this, carbon microphones were used until electronic amplification permitted the use of capacitor microphones and dynamic microphones, which are not plagued by noise. Crystal and electret microphones are used at the present time, but they are possible only because of electronic amplification by solid state devices. The sensitivity of the telephone carbon microphone was about -15 dB re 1 V/μbar, and its internal impedance was 30 to 100Ω.
The carbon microphone is subject to several diseases, mostly caused by damage to the carbon granules. They can "pack," greatly reducing the resistance, and the output, of the microphone. This can happen when too large a current is passed through the microphone, causing the granules to heat and fuse. If the circuit to a carbon microphone is opened while current is flowing, and the circuit contains an inductive element, the inductive surge can damage the granules. This doesn't happen, of course, when a microphone plug is simply pulled out of a jack, since there is no inductive kick.
Electret
condenser microphones (ECM) are generally used at the present time for
telephones and similar apparatus. The arrangement of these microphones is shown
at the right. The electret is a permanently polarized polymer film, the
electrostatic analogue of a magnet, that creates an electric field between the
diaphragm and the back plate, establishing an electric field between them whose
strength is independent of the spacing. This eliminates the necessity for a high
bias voltage, a very great convenience. The voltage across this capacitor, which
changes when the diaphragm is moved, is applied to the gate of an FET, located
inside the transducer. The output voltage is developed across the load resistor
RL. A typical ECM has a frequency range of 50 Hz to 13 kHz, and a
sensitivity of -60 dB re 1 V/μbar when working into a load resistance of 1 kΩ.
The power supply is VS = 1.5 - 10V, with an optimum value of 4.5 V.
The collector current of the FET is about 0.5 mA. The maximum SPL is 120 dB re 2
x 10-4 μbar. These are only typical specifications, of course. I do
not know how neutralization of the electret by charge pickup is prevented.
Indeed, electrets are used to attract charged dust. Magnets do not suffer from
this, because there are no free magnetic charges.
The
telephone receiver used with the transmitter discussed above in a typical desk
set is shown schematically at the left. There was plenty of room for the large
permanent magnet made of hard steel. Great improvements in permanent magnet
materials have allowed the receiver to be reduced greatly in size. The smaller
receiver made possible by Alnico magnets, together with the more compact
transmitter, were combined in the convenient handset in the later 1930's. Just
as in the case of the transmitter, telephone receivers have always used the same
principles, from the first receiver designed by Bell to the present.
Telephone-type receivers are also used in headsets. Of course, modern
developments have provided alternatives.
The iron diaphragm is attracted by the permanent magnet. Coils are mounted on soft-iron pole pieces, and the windings are arranged so that the total magnetic field is strengthened or weakened in time with the current in the coils. This urges the diaphragm to move in step with the current. If the diaphragm were not biased by the permanent magnet, it would move at twice the frequency of the voice currents, which is certainly not desired. In the diagram, the permanent magnet slides into the hard-rubber shell, while the cap is screwed onto it, which holds the diaphragm in place. In the early receivers, the diaphragm resonated at about 1000 Hz, which was desirable since this strengthened the most important speech frequencies. Through careful development, the response was flattened between 300 and 3000 Hz. Combined with the new transmitter, the frequency response became quite uniform, providing good fidelity and eliminating some less-desirable peculiarities of the previous telephone sound. The present-day receiver is usually a piezoelectric audio transducer (not known for audio quality) driven by an integrated-circuit amplifier.
A
transmitter, battery and receiver can be connected in series to make a simple
telephone circuit. When two-way or duplex communication is desired, it is better
to connect the transmitters and receivers in parallel, one of each at each end.
However, the low impedance of the transmitters, only 30 to 50Ω, seriously shunts
the circuit. This problem was solved by Émil Berliner through connecting the
transmitter through an impedance-matching transformer, which was called an
"inductor." Now the speaking current did not pass through the line, which was an
advantage for the receivers, since they should not have a DC current passing
through them. With the ringer and magneto for producing the signalling current,
all the main elements of the telephone were now present, as shown in the diagram
at the right.
A few other components were still necessary. A ringer and a magneto for signalling were added, with a switch that would disconnect the ringer when the magneto was operated, with a capacitor to block the low-frequency ringer current from the receiver and transmitter. To conserve the battery, the receiver was hung on a lever switch that disconnected the battery when the receiver was "on-hook" or not in use. These elements were usually enclosed in a wall box, with the transmitter on the front and the receiver hook and magneto handle on the right side. The ringer gongs were perhaps on the top of the box, and the batteries, usually two #6 dry cells, were inside behind a door.
When common-battery switchboards were used, the magneto could be dispensed with, as well as the dry cells. The lever switch was connected so that when the receiver was off-hook, a signalling current was allowed to pass through the telephone. Then, most of the components could be put in a small wall box, connected with a desk set through a short cable. The desk set was a support for the transmitter, with a receiver hook on one side. It was generally held in the left hand (or left resting on the desk) while the receiver was held to the ear with the right hand. This was a very convenient arrangement that was used for many years. When dialing was introduced, the dial was on the base of the desk set. Further miniaturization meant that all the equipment could be put in a small box with a handset cradle on the top. The wall box was eliminated, and the handset could be held to mouth and ear at the same time.
Now let's discuss in detail a rather different kind of telephone set, the Army field telephone. The classic EE-8 was much like a home telephone, but was arranged for field use and had some interesting features.
The EE-8 telephone came in a heavy leather or canvas case, and weighed 16.5 lb. The TS-9 handset was on one side, the body of the set on the other side, with the crank handle of the magneto, which folded out, accessible outside the case. The line was connected at two large binding posts on the top of the body, marked L1 and L2. A third, smaller, binding post connected to the (-) terminal of the battery. The three wires of the handset cord connected with binding posts marked R for receiver, C for common, and T & B+ for the transmitter and (+) terminal of the battery. The battery terminals could be used for connecting an external battery. Speaking current was provided by two 1.5 V BA-30's, the Army's carbon-zinc D-size dry cell, the usual flashlight battery. There was a press-to-talk switch on the handset that was solely used to conserve batteries, not to change a link from receive to transmit.
The
circuit of the EE-8, as arranged for normal local-battery operation, is shown at
the right. T is the carbon transmitter, R the receiver. B is the battery, and S1
is the handset switch. L1, L2 and L3 are on a common core, and constitute an
autotransformer. When the handset switch is pressed, talking into the
transmitter creates a voice-frequency signal that is stepped up by a turns ratio
of about 10:1 and applied to the line. The autotransformer, traditionally called
an "induction coil" and first used by Berliner, matches the low impedance of the
transmitter, about 27Ω, to the line. C1 blocks any direct current from passing
through this part of the set. The receiver is connected across L3 and C2, which
form an anti-sidetone circuit. The components are chosen to reduce the
signal heard in R when transmitting, but to match the line to the 256&Ohm;
impedance of the receiver. Note that DC cannot flow through the receiver, since
it is blocked by C1 and C2. The speaking current of T is about 100 mA, somewhat
greater than that of civilian telephones, to ensure a good output level. The
Army says that L1 is 120 mH and 8-22Ω, L2 is 5 mH and 3 Ω, and L3 is 80 mH and
370Ω, and that the inductance from L2 to the line is 42 mH, but I do not
understand exactly what these figures mean.
The sidetone is what you hear in the receiver when you yourself are talking. If it is too loud, you will speak softly and not be understood. If it is too feeble, you will shout into the transmitter. The circuit in the EE-8 is the one devised by G. A. Campbell in 1918 that improved on the simple transformer of Berliner. Because I do not know the parameters accurately, I have not analyzed the circuit. Let's assume that it works as intended, attenuating the signal from the transmitter and using as much of the signal power from the line as possible.
Signalling in this connection is by means of low-frequency alternating currents generated by magneto G, called a GN-38. A GN-38 produces 16 Hz at about 100 V when its handle is turned, and has a resistance of about 400Ω. Most magnetos have rotating permanent magnets and fixed coils in which the AC is induced. A switch operated by holding in the magneto handle connects the magneto with the line only when it is operated. Of course the ringing current can be heard in the receiver, but C1 does a lot to block the low frequency. An incoming ringing current is also blocked by C1, as well as by the high impedance of the transmitter-receiver part of the set, so most goes into the ringer A, called an MC-131. This is the traditional polarized ringer invented by T. Watson, Bell's assistant, in 1878 that moves a clapper at the frequency of the ringing signal. In the EE-8, there is a single gong and a clapper that moves inside it, hitting two stops on the gong. In early models, an attempt to save money was made by constructing the gong and clapper of steel. The problem with this was that stray magnetism made the clapper stick to the gong! If you could find a brass gong or a brass clapper, then a simple replacement solved the problem. In some military situations, a phone ringing can be somewhat undesirable. The magneto handle can be held in during an emergency, but more permanently the gong could be removed, and only a dull vibration was then produced. Friction tape could also be put where the clapper struck. If the ringer was completely disconnected, a visual signal was available, the U-4/GT, which was simply a neon lamp that flickered when the ringing current was present.
In the local-battery arrangement, the ringer is constantly across the line. Its DC resistance is only 1300Ω, but its inductance is 3H, so it offers high impedance to voice currents at 1 kHz (19 kΩ). Note how signalling and voice currents are separated, and also that DC does not appear on the line. The telephone set contains only passive components, if the carbon transmitter is considered a passive component since it produces no electrical gain, and is merely a transducer. The transmitter in the EE-8 has an average output of 3 mW.
To use the telephone, rotate the magneto handle a few turns to alert the operator at the switchboard, or the other telephones on the line if there is no switchboard. Pick up the handset and tell the operator the desired connection. The operator will make the connection and ring the other party, which you should hear. When the other party answers, carry out the conversation. At the end of the conversation, replace your handset and give the magneto a few turns. This is called "ringing off," which will signal the operator that the connection can be cleared. This is a very simple and adequate procedure.
Civilian
telephones are almost always operated from a common-battery switchboard,
so that the central office supplies the DC talking current, releasing the
subscribers from the bother of maintaining their batteries (see below). The DC
that is supplied is also used for signalling, so that magnetos are not required.
The EE-8 uses only local battery talking current, but can be arranged so that it
acts properly with common-battery signalling. The modified circuit is shown at
the left. It is selected by rotating a switch with a screwdriver from the "LB"
or local-battery position to the "CB" or common-battery position. What this does
is take a shunt off C3, so that DC cannot pass through the ringer, and make a
circuit through the holding coil L4 and the lever switch operated by
picking up the handset. When the lever switch is held down, the normal position,
the set presents an open circuit to the line, and no current flows. When the
handset is lifted, DC can pass through the 100Ω of the holding coil. This
normally establishes a current of 240 mA or so that signals the operator that a
connection is desired. The inductance of the holding coil is 1H, so it presents
a large impedance to voice signals and so, like the ringer, does not shunt the
circuit excessively.
When the EE-8 is used with a common-battery switchboard, it works just like the phone at home, except that the talking switch must be depressed, and batteries must be installed. The usual D cells, 1-11/32" diameter by 2-13/16" long, when used 2 hours a day at a drain of 100 mA will fall to 1.0V in about 103 hours, so that one pair of batteries will last about 51 days at this level of use. For 4 hours per day (which is a lot of talking) battery life is only 36 hours, or 8.5 days. Under continuous use, the batteries will survive only 18.5 hours. The batteries used with traditional magneto telephones were the large #6 dry cells. At 2 hours per day, they will last 400 hours, or 200 days. Replacing them twice a year was about all that was required.
On a traditional local-battery magneto telephone, operating the lever switch closed the transmitter circuit. In this case, the batteries were under load for the full duration of the conversation. The handset switch of the EE-8 halved the load on the batteries if each party talked equally, and so doubled the life of the batteries. This, then, was a very worthwhile feature, even if not tolerated in a civilian telephone. It made the use of two small D cells, instead of two large #6 cells, practical.
EE-8 telephones could be used at the ends of a point-to-point circuit, which offered not only convenience, but also the longest range. Using W-110-B field wire, a tough twisted-pair wire with mixed copper and steel strands, laid on the ground, a range of 11 miles could be expected. On a line of #12 copper wire on poles with a ground return, the range could be 230 miles. When connected to a switchboard that used trunk circuits, the local loop should be no longer than about 2 miles with W-110-B, or 46 miles with open wire. These ranges are quite creditable, considering that no electronic amplification is involved, and no power other than the two D cells in each set. Similar telephones were very useful in rural areas for many years.
More than the two sets at either end could be bridged across the same line if desired, but the range was shortened, of course. The principal problem was not in speaking range, but in ringing range. Each ringer is a heavy load on the magneto output, and too many ringers would enfeeble them all. However, such "party" lines greatly reduce line expense, and were often used. Each set was assigned a different ringing code, such as 1, 2, 3 and so on. This is quite different from the case in DC telegraphy, when any number of offices can be on the same line with no loss of efficiency, so long as sufficient battery was applied.
With more than a few subscribers, however, point-to-point circuits are hopelessly extravangant, and some means had to be found to connect any two sets. The Army had a simple system that demonstrated all the principles. At the "switchboard" each line terminated in a plug-jack with two contacts, connected by a neon lamp and resistor. To connect two circuits, all that was necessary was to plug one plug-jack into another. When the magneto was used to call the operator, the neon lamp for that circuit glowed. The operator would then plug his own circuit in, and receive the desired connection. He would then plug his set into the plug-jack for the desired connection, and operate his magneto. When the party answered, he would connect the two circuits, and remove his own plug. At the end of the conversation, the flashing of the neon when the caller rang off would show that the connection could then be cleared.
The arrangment used to make temporary connections between telephone lines from the early days was called a plug and jack. A "jack" was so-called because it was adapted from an earlier device used to make temporary connections for the telegraph, that looked to some observers like a jackknife, and "jackknife connector" was shortened to "jack." The telephone jack was completely different, however, with a cylindrical coaxial plug that engaged spring contacts when inserted in the jack. The cylindrical contact was called the "sleeve," and the central contact the "tip." For telephone use, a third contact, insulated from the other two, was arranged between the tip and the sleeve, and called the "ring" from its shape. In moving the spring contacts, switches could be open or closed as in a relay.
The first commercial switchboard was installed at New Haven, Connecticut in 1878, with 8 lines and 21 subscribers. A switchboard had been used by a burglar-alarm company (one of the first uses of the telephone) in Boston the year before. We shall now discuss with some thoroughness the manual switchboard, which has largely disappeared, but is a very interesting technology, able to handle a surprisingly large traffic with simple apparatus. A basic manual switchboard for magneto telephones was provided with magnetically-released drops so that the operator would not have to watch constantly for the transient indication of a neon lamp or anything similar, and the calling line would be persistently identified. The drop was near the jack for the corresponding circuit. It was found best to make three contacts with the plug and jack for a circuit. These were called tip (T), ring (R) and sleeve (S), as we have already mentioned. The subscriber line was connected to the tip and ring contacts, while the sleeve was grounded or used for special purposes. So far as the telephone set is concerned, it does not matter which line wire is T and which R. A party desiring a connection would call the operator by several turns on the magneto, and the ringing current would release a drop at the switchboard. When a drop fell, the operator would plug the answering end of a patch cord into the jack. Then, he would manually restore the drop, or it could be restored automatically for him. [I shall call operators "he," although women made excellent operators and dominated the profession.] By operating a switch, his own set could be connected to the calling circuit, and the desired connection could be requested. The other end of the patch cord was then plugged into the jack for the called party, if it was not already occupied. A ringing switch then disconnected the calling party, and connected the ringer, which could be a magneto. When the called party answered, the connection was complete, and the conversation could take place. When the caller, or the called party, rang off, the drop would again fall, and the operator could restore the circuits. A more elaborate local-battery switchboard cleared the calling drops automatically by a voltage applied to the sleeve connection, and had a drop in the cord circuit that was operated when the caller or callee rang off.
When there were a large number of circuits, two operators could work a board divided into two sections. Each operator had the jacks for half the lines before him, and was assigned to answering them. To complete a call, an operator could reach the jacks in both boards without inconvenience. Several thousand lines could be managed successfully in this way, with perhaps twenty or thirty simultaneous conversations when necessary.
To relieve the subscriber from the necessity of maintaining batteries, the speaking current was supplied over the subscriber's loop from a battery at the central office, the common battery. This kind of switchboard was invented by J. J. Carty of the New York Telephone Company, and the first was installed at Lexington, MA in 1893. The same common battery served all the telephones connected to the switchboard, so it was necessary to isolate the voice currents from the low impedance of the battery. A separate battery or a tap supplied a supervisory voltage to detect when a subscriber went off-hook by the current that then flowed, since the magneto was usually eliminated at the same time. This current operated a relay that lighted a line light that served the same purpose as the earlier drop. When the operator answered the call by inserting the plug on a connection cord, this light went out. As before, the operator connected his set and learned what connection was required verbally. When the other end of the cord was plugged into the called jack, a supervisory light went on showing there was no current in the called circuit, since it was still on-hook. The called party was then rung by operating the ringing switch, which connected the ringing machine to the called line, instead of having the operator operate a magneto. When the called party answered, the supervisory light went out, showing that the connection was complete. When either party "hung up" the cession of the signalling current lighted one or both of the supervisory lamps, showing that the connection could be cleared. When a call was in progress, no lights were shown. A light always meant that something needed attention. The circuits for doing this will now be described. There have been many variations, but the principles are the same.
A
typical jack circuit for a multiple switchboard is shown in the diagram. This
circuit not only has to bring the subscriber loop wires to the switchboard, but
has to indicate when the subscriber goes off-hook to request a connection. To do
this, a battery B supplies a signalling current to the tip and ring lines of the
loop. When the subscriber set is on-hook, the set presents an open circuit to
the line, and no current flows. The line relay LR is released, and the line
light LL is dark. When the subscriber goes off-hook, a small current flows that
is sufficient to pull in the line relay, and LL lights, notifying the operator
that attention is requested. When the operator plugs an answering plug into the
jack, the sleeve connection supplies current that activates the cutoff relay CR,
opening the circuit from battery B to the subscriber loop and the line light. LL
then goes off, and speaking current is supplied by the cord circuit of the
switchboard. The jack may have switches that disconnect the signalling circuit
when the plug is inserted. In this case, CR is unnecessary, and the sleeve
connection need not be used. However, if there are multiple jacks, then the
sleeve connection is necessary to signal that a circuit is in use.
A typical
cord circuit for a multiple switchboard is shown at the left. A number of these
cords are in front of the operator, with which he answers calls with the
answering cord when a line light lights, and completes calls with the completion
cord, connecting the line jacks of the two subscribers. The circuit may appear
complicated, but it is actually fairly simple if broken down into modules. One
function is to supply speaking current to both subscriber loops into which it is
plugged. This current comes from the battery B1 through the repeating coils T1
and T2. It is necessary to isolate the voice signals from the battery so they
are not shunted by its low resistance. This shows one way of doing this, and
probably the best. Other methods will be discussed later. The operator's set can
be connected to the circuit by operating the listening switch LS, and ringer
current can be put on the completion end with ringing switch RS. When RS is
operated, the answering side is isolated from the ringing current to avoid an
unpleasant sound. With automatic switching, a mild signal would be put on the
answering side to show that the called party is being rung, or a distinctive
busy signal. A live operator, however, would probably advise the caller that the
ringing is continuing, or that the called line is busy, by voice. Two ordinary
switches are shown, but these functions can be combined in one lever switch that
connects the operator's set when moved one way, and rings when moved the
other.
The two supervisory circuits at the bottom advise the operator of conditions on the subscriber loops. Whenever speaking current is flowing in one of the loops, the supervisory relay SR is held closed, and the supervisory lamps SL are shunted, and do not light. The sleeve connection completes this circuit through ground and the cutoff relay (the voltage drop across this relay is what signals the operator that the line is busy when the sleeve is touched by the tip of the answering plug). Whenever either party goes on-hook, the corresponding supervisory lamp lights. This indicates to the operator that the line is no longer required, and the connection can be cleared. When a conversation is in progress, no lights are lighted, and the existence of the connection is shown by the cord's being in use. When a supervisory light goes on, the operator may clear the connection. When a line light goes on, the operator answers the request. A light, therefore, is a sign that immediate action is required.
The usual common battery was 24 V or 48 V, mostly the latter. The voltage had to be high enough to supply sufficient talking current on the most difficult subscriber loop. For signalling purposes, 24 V was often adequate. The positive battery terminal is connected to the tip and is also grounded. The negative battery terminal is connected to the ring. Of course, all kinds of variations and exceptions may exist. What you measure on an on-hook line is the supervisory voltage, which may or may not be the same as the DC voltage supplied on an off-hook line. The speaking current is now used to feed the electronics in the telephone, instead of passing through carbon granules. The T-R voltage can be from -47 to -105 V, the speaking current from 20 to 120 mA, typically 20 to 80 mA, with a subscriber loop resistance of 1300Ω maximum. The standard ringing signal is 20Hz, 90V, but can be between 16 and 50 Hz, 40 to 130 V rms. Any polarity-sensitive apparatus must expect the T-R voltage to be of either sign, so a bridge rectifier is needed. The voice signal passband is from 300 Hz to 3 kHz. The original passband was determined by the apparatus, and was approximately in this range. With the use of carrier equipment, it was necessary to enforce the passband rigidly to prevent aliasing.
Subscriber telephone connections are now made with 4-contact modular jacks that replaced the older 4-pin connectors. Two-pair cables are used, usually with #24 AWG wire, which are color-coded green, red, black and yellow. Nevertheless, the phone still requires only two wires, and tip (+, grounded) is assigned to the green, ring (-) to the red. The other two wires are there mainly for use in case of trouble in the red and green wires. They can be used for a second line, in which case tip should be black and ring yellow. They are also commonly used to supply low-voltage AC for lights in the telephone set, when there is an accessory transformer nearby. The tip and ring connections can usually be interchanged without any problem, though some tone dialing only works one way (the dial tone does not disappear when connected the other way). A line connector uses L1 for tip, L2 for ring. Wiring your own phones is dead easy, but take a phone off the hook so you will not be rung while working. Ringing voltages can ruin your day. Up to five phones can be paralleled across your subscriber loop and they will still ring.
Two
additional methods of connecting a common battery are shown at the right. The
main idea is preventing the low impedance of the battery from shunting the
lines, and this can be done with inductors. Two inductors are required, since
the tip side of the battery is grounded. If only one inductor were used, the
line would become unbalanced, with one side grounded to AC as well as DC.
Circuit (a) can be used if the subscriber loops are all of about the same
resistance. If they are not, loops of low resistance will drag down the current
in those of high resistance because of the resistance of the inductors. Circuit
(b) solves this problem, using capacitors to connect the subscriber loops. The
circuit given above using repeater transformers is similar, except for the
coupling.
One operator can only handle a limited number of lines. When there are more lines than this, they must be divided among several switchboards. Any one operator can complete connections only among the lines on the same keyboard. If a line is requested that is assigned to a different keyboard, then the operator must communicate with that keyboard through a trunk, and ask the operator there to complete the call from the trunk. This works, but is slow and requires a large number of point-to-point trunks. On a multiple switchboard, any operator only answers a limited number of lines, but can make a connection to any line connected with the exchange. This is done by providing multiple jacks so that at least one is within the reach of any operator. Since an operator was considered to be able to reach a jack in his own board as well as those in the boards on either side, this requires that jacks be repeated every three board sections. The operator usually had before him the answering jacks, or incoming lines, at the lowest level where they were easily monitored. Above them were jacks for trunk lines for transferring calls to other central offices, and on the top, in the largest section, were the multiple jacks for completing calls. Right in front of the operator was the row of patch cords, with supervisory lights, listening and ringing switches. It is not difficult to conceive how all this worked in practice, and it usually worked very well. In very large exchanges, operators often worked in pairs on multiple boards, one answering calls and the other completing them. All incoming calls were "trunked" by the answering, or "A" operator to the appropriate board for completion. For every board, there was a trunk to its "B" operator, who would complete the call.
When completing a call on a multiple board, it had to be determined that the line to be used was not already connected. On non-multiple boards, this was easy to determine because there would be a plug already in the jack. We have already explained how the sleeve could be used to indicate when a line was busy. When a line was not busy, the sleeves were grounded. If the tip of a plug was touched to the sleeve of the jack intended for a connection, a click would be heard if the line was busy. If it was free, both tip and sleeve were grounded, and there would be no click. If an error were made in doing this, an unintended conference call was created.
In those days, suppose we allowed a maximum of 10,000 lines per central office or exchange, which could be managed with a multiple switchboard. The exchange was given a name, such as PEnnsylvania, and a number, such as PE-5000 (possibly information at Penn Station, New York City) was requested orally as "Pennsylvania 5000." If the call came from SUnset 2027, then the Sunset operator answering it would trunk the call to Pennsylvania, saying "Sunset 6500" to show what exchange originated the call, and where it was going, and an operator there would complete it. If the call came from PE-1018, then the Pennsylvania operator who answered it could also complete it immediately. A wonderfully large number of lines could be handled in this manner. The operator would understand just the letters, but using the whole exchange name reduced misunderstandings, and made the number easier to remember. In Denver, the M. L. Foss engineering supply company could be reached at KEystone 5151 in 1940, a number which will probably stick in your mind with ease, unlike 213-555-9763, which will immediately be forgotten. This attractive system echoed down the years, only adding a number to the exchange, like PE6-5000 when machine switching, or "dial phones" came in, and the number of lines increased. Finally, the alphabetical names were dropped and replaced by numbers, though the letters have survived on the dialling buttons, and are now used for other purposes, such as CALL ATT. In Denver, we lost SKyline for a prosaic 73. Now 10 digits have to be dialled, and there is no easy way to remember them, so apparently many people just dial and see what happens (most of my calls are wrong numbers). It's clear that with the end of manual switchboards there was no further advantage in exchange names, and they became simply an expensive ornament.
Calls completed outside a local calling area were called toll calls, for which there was an extra charge depending on distance and time. The operators handling them had to keep toll tickets showing the time and routing of the call for billing purposes. These long distance calls were still handled manually by operators when practially all local calls were mechanically switched, as late as the 1960's. In the ringdown method, the originating operator calls another operator at the destination, or to an intermediate point on the route. The call is then transferred to this operator, who completes it by dialling the called party, or by passing it on again. Mechanical switching of long-distance trunks made this much easier, assigning area codes to calling areas, and finally the subscriber could complete the call fully mechanically, using the area code in front of the subscriber's local number. The average time to complete a long-distance call was 15 minutes in 1918, 5 minutes in 1928, and less than 2 seconds in 1938 (or so the Bell System said). A large part of this improvement is simply in reducing the number of connections required.
Telephones are inherently inefficient because a dedicated circuit connection must be established between the ends of the link, and communication occurs at the leisurely speed of speech in a rather wide bandwidth. Only a rather modest amount of information can be exchanged in say, five minutes. The telegraph, however, notably as realized in modern internet communications, is packet switched, and a packet containing a considerable amount of information requires very little expense. An e-mail message is almost free, compared to the cost of transferring the same information by telephone. We have come full circle here, and now the telegraph reaches into most homes and offices. It is strange that the public telegraph service died, perhaps due to poor marketing and absurd pricing, just as the new telegraph was born.
One curious bit of telephone lore can be recalled here. Years ago in Britain, public pay telephones in the red phone boxes and elsewhere were black metal boxes with a handset and a dial. When the handset was picked up, nothing was heard until you put some money in the coin slot. When you did this, you could get a dial tone and dial your party. When the party answered, you spoke in vain until you pressed a button labelled with a large "A" in a circle. Then you heard your money drop into the money box where it was forever gone. If your party did not answer, or it was the wrong party, you pressed a button labelled with a large "B" in a square, and your money came tumbling back. If there was a manual switchboard, you could talk to the operator and request your number without putting in any money. She would tell you how much the call would cost. A penny made a dong, a sixpence a ding, and a shilling a ting-ting, so the operator could tell how much you inserted. If it was the correct amount, you were connected, and warned later to put in more when your time was running out. Now you can get a dial tone when you lift the receiver, which means the phone can be used in emergencies without money. When your party answers, a persistent beeping can only be stopped by putting in the correct change.
I have not been able to find EE-8 telephones for sale, and wonder where they all went. I would very much like to have a pair for demonstration purposes.
A. L. Albert, Electrical Communication, 2nd ed. (New York: John Wiley & Sons, 1940). Chapter VII.
TM 11-333, Telephones EE-8, EE-8-A and EE-8-B (Washington, DC: War Department, 1945).
J. L. Fike and G. E. Friend, Understanding Telephone Electronics (Fort Worth, TX: Texas Instruments, 1983). Distributed by Radio Shack, but RS does not appear to still sell books.
A more detailed history of the telephone can be found on the web at Galaxy Phones. Cookies can be rejected without penalty here.
Composed by J. B. Calvert
Created 23 August 2003
Last
revised 7 September 2003