Teletypewriter

Before the Model 33 and ASCII code


Introduction

Some of the earliest telegraph instruments were printing telegraphs, such as the Hughes machine, or indicated letters directly, such as the Wheatstone ABC system. However, they were either slow, or else were complicated and temperamental mechanisms requiring good circuits and constant skilled maintenance. The fundamental idea in all such devices is the maintenance of synchronism between the transmitting and receiving mechanisms. This could be provided by two wires (as in Edison stock tickers), but for practical telegraphy single circuits only were practical.

The breakthrough was made by Baudot in 1874, who used a five-element code transmitted by a sort of piano keyboard and received by a synchronous mechanism (the distributor) that selected the letter to print. This mechanism was refined into the teletype machine that dominated telegraphy in the first half of the twentieth century. The Baudot system first spread through France, and became general worldwide by 1910. Today's Internet is a telegraph network, made possible by semiconductor electronics and other developments, that, together with a digitized telephone network, provides the majority of today's telecommunications. This page will describe the technical features of the classic mechanical teletype apparatus used in the 1950's. General history and later developments will be found elsewhere.

The advantages of a printing telegraph are speed, accuracy, permanence of record and ease of relaying through the use of paper tape. The greatest advantage, however, is in not requiring skilled operators. A teletype sending operator need only know how to type, and a receiving operator only how to tear off the paper from the printer, in distinction to the highly-skilled, highly-paid and highly organized Morse operators. This is the outstanding reason for the dominance of the teletype. The disadvantages are weight and size, power requirements, expensive mechanical maintenance, and the necessity for good circuits, in addition to considerable noise. The modern application of electronics and the elimination of nearly all mechanism overcomes all of these disadvantages. The inkjet printer is much cheaper, lighter and quieter than a mechanical teleprinter!

Teletype used electricity only for signalling. The transmitter opened and closed contacts at specified times that controlled the line, while the receiver used an electromagnet driven by the line to select mechanical settings for printing. Everything else was mechanical. This elaborate and advanced mechanical technology was also seen in mechanical calculators of the time, in watches and engine governors, and in other delicate mechanisms that have now been completely superseded by electronics.

How a Teletypwriter Worked

The U.S. Army's standard teletypewriter of the 1950's weighed about 200 pounds. It was 42-1/8" high, 18" wide, and 21-5/16" deep, composed of a keyboard unit, a typing unit, a motor unit, and a base unit, all on a metal table with everything painted green. The keyboard had the keys arranged as on a standard QWERTY typewriter, but in only three rows with 31 keys and a spacebar. There were only capital letters. Keys were provided for LTRS and FIGS shift modes. Each key had the LTRS function below, and the FIGS function above. The top row, for example, had the digits from 1 to 0 above the letters Q to P. Paper, usually rough absorbent yellow paper, was supplied from a roll at the back of the printing unit, and after passing around the platen of the printer, exited at the top, where it could be torn off above the window allowing a view of the printing. The teletypewriter is shown without its cover at the right to give an impression of its mechanical complexity.

The standard speed of transmission of U.S. teletypes was 60 6-character words per minute. This was about twice as fast as a skilled manual operator could send or receive. The transmitting shaft of the teletype ran at such a speed that each revolution corresponded to one character, so the basic speed was 360 rpm. The actual speed was 368.1 rpm, which allowed time for the mechanism driven through a clutch to stop and start for each character. In the U.K., the standard speed was 66 wpm and 404 rpm. The drive motor, geared to the transmitting and receiving shafts through clutches, ran continuously at 2100 rpm, and was closely governed to the desired speed. A tuning fork vibrating at 87.6 Hz was used to check the motor speed by viewing a pattern on the rim of the flywheel through a slit between the tuning fork tines, in the manner of a stroboscope. This is a good example of the mechanical solution of problems that would now be attacked electronically. If there were 25 spots on the flywheel, 35 rps x 25 = 875, so that about five spots would pass in one half-vibration of the tuning fork.

The telegraph line was said to have two states, marking when the circuit was closed, and spacing when it was open, names derived from the original Morse register system. A teletype circuit could be neutral, meaning off-on and independent of polarity (direction of current flow), or else polar, when the two states corresponded to opposite directions of current flow. Practically all commercial circuits were polar, but the U.S. Army used neutral circuits. The modern RS-232 asynchronous communication standard uses polar line circuits. Two teletypewriters were connected in series at each end of the line, so that the same current passed through both, and the line was normally held in a marking state. This gave a half-duplex link, in which either end can transmit, but not at the same time. RS-232, with which you may be familiar from computers, is full duplex, but, of course, requires two wires for this.

The U.S. standard Baudot teletypwriter alphabet is shown at the right. The rightmost bit is transmitted first, the leftmost last. A "1" is mark, a "0" space. The six codes at the bottom of the table are independent of LTRS or FIGS shift, having the same interpretations in either. A special alphabet was devised for weather telegraphs, since weather networks had their own special requirements. This alphabet developed into the ASCII 7-bit alphabet without shifts, since electronic equipment do not need them. The idea, however, is exactly the same.

Transmission is easiest to understand. When a key was pressed, its bottom surface pressed inclined cam surfaces on each of five selector bars so that the bars moved to the left for a mark, and to the right for a space. Further motion then moved a bar that closed the transmitting clutch. Then the transmitting shaft began to rotate, carrying with it six circular cams for the start bit and each of the five data bits. The six pairs of contacts were in parallel across the line. Each cam held its contacts open, allowing them to close during its time slot if they were not held back by a locking bar driven by a selector bar. The start-stop cam normally held the line closed. As the shaft began to rotate, the circuit was opened, which would start the receiver at the other end of the circuit. Then each of the five data bits occupied their time slots. If the selector bar was moved to the right, the contacts were held back so they could not follow the cam, and remained open. Otherwise, the contacts closed. When the shaft had rotated almost one full revolution, the start-stop contacts again closed for an interval of about 1-1/2 data bit times before the shaft came to rest, triggering the clutch to disengage and the transmission shaft to brake to a halt.

Receiving is much more complicated, and only the basic idea can be sketched here. The receiver had a selector magnet energized by the line current. When it dropped out as a start bit (spacing) was received, a clutch was engaged and the receiving shaft began to rotate with the standard speed. The selector magnet then was energized or released during each bit time, which moved selector vanes accordingly. When the printing bail, driven by the motor, was moved, it operated only the selected type head, driving it to print as in a typewriter. There was also a function bail that did whatever else could be done besides printing, such as carriage return, line feed, letters shift and numbers shift. The most interesting part of the receiver from an electrical standpoint was the rangefinder, an adjustment made through a small door on the left of the printer. This adjustment selected the window for the mark-to-space and space-to-mark transitions. With the operator at the other end sending RY signals (these were codes 01010 and 10101), the rangefinder was moved one way until the reception became garbled, and this was repeated moving the other way. A setting halfway between was optimum. This adjustment compensated for distortion in the received signal.

Teletype Circuits

Teletype circuits may be neutral or polar d.c. (direct current), carrier, or radioteletype. An example of the carrier type is the modem link for moving data over a telephone line. A more elaborate example is a microwave relay link, which can multiplex many telegraph circuits on a single UHF carrier. Radioteletype is really a kind of carrier, but a special one that uses a HF link and may be very-long-distance. Microwave relay and HF radioteletype were both available and widely used in the 1950's, but satellite communications and optical fibre networks were not yet in service. I will discuss mainly the traditional d.c. circuits here.

A circuit is called metallic if both conductors are of wire, and ground return if only one conductor is of wire, and the other is formed by a ground connection. The ground return was discovered by Steinheil in 1838, and was a general feature of telegraph circuits from then on. The earth is a very good conductor, and the use of a ground return cuts the resistance of a telegraph circuit in half. What is also of importance to commercial enterprises is that it cuts the wire line cost in half at the same time. Telephone circuits do poorly with ground return circuits, but telegraph circuits can function rather well in many circumstances.

A typical teletypewriter circuit had a current of about 60 mA, and was supplied by 110 VDC sources. Resistances were usually used to adjust the current to the proper value. For short links, 20 mA was suitable, and became a standard.

It is necessary to take a great deal of care to get a good ground return. Unpainted, clean metallic pipes that travel distances underground are a good choice when they exist. If one is not available, a ground has to be made. Usually, a rod is pounded into the ground so that it is in contact with soil moisture below frost level. A ground rod embedded 2 ft may have a resistance of 90Ω, 4 ft 50Ω, and 7 ft 30Ω. The resistance figures only show the effect of deeper penetration. Actual resistances depend on the type of soil. The diameter of the ground rod has little importance. A shallow pit 6" deep can be dug, and the rod pounded in until only about 3" is above ground, to which connection is made with a good clamp. Then the pit is filled with 5 lbs of coarse salt, and water poured on. The ground resistance should decrease as the salty water trickles in. Dry sand, gravel and rocks are impossible to make a good ground in. Rods at least 10 ft apart can be connected in parallel to improve the ground. Ground resistance can be measured by using a line from another station with a good ground and measuring the voltage with a known current (or the current with a known voltage).

A bad ground will cause numerous symptoms. If the ground resistance is too high, it may be impossible to get the required line current with the voltages available. There will be great variations between wet and dry weather that will necessitiate adjustments. If more than one circuit shares the same or neighboring ground rods, there may be crossfire (interference or crosstalk) between different circuits due to coupling in the common ground resistance. Ground return circuits suffer from line leakage (to ground), wet-dry effects (because of insulation resistance and leakage), and induction from power lines, neighboring circuits, and earth currents due to ionospheric effects. All of these disturbances are minimized with a metallic circuit.

A simple teletype circuit is shown at the left. A teletypewriter is represented by its transmitter contacts and its receiver electromagnet. This is exactly like a manual telegraph circuit with key and sounder. The divided battery is a means to lower leakage by keeping the average line voltage as small as possible. Here, one end of the line is at +55 V with respect to earth, while the other is -55 V with respect to earth, so that the average absolute potential difference is half what it would be with a battery at one end of the line only. On the other hand, it is often more economical to keep all the batteries at a central location. The total resistance of line and instruments for a current of 60 mA is 1833Ω with 110 V of battery, neglecting leakage.

A common way to get a good telegraph circuit when a telephone circuit was already available was the simplex the telegraph circuit on the metallic telephone circuit using repeating coils. These were 1:1 transformers with center taps. Teletypewriters could be connected in a normal ground-return circuit between the center taps at the two ends. They would not disturb the telephone circuit, while the telegraph circuit would have the two line wires in parallel, decreasing its resistance. The circuit is shown at the right. The proper operation of this circuit, and all like it, depended on balance between its two sides. Lack of balance would cause interference between the telephone circuit and the telegraph circuit. The line wires had to be transposed at suitable intervals to reduce inductive crosstalk. Nevertheless, the telegraph circuit was available with little extra expense.

Two metallic telephone circuits could provide a metallic telegraph circuit in addition with no extra conductors. This was called a phantom group, which again depended critically on the balance between its components. A phantom teletype circuit is shown at the left. Still another arrangement, the simplexed phantom group, could support 3 telephone circuits and 1 teletype circuit over only four wires. In this case, the teletype circuit is simplexed on the phantom telephone circuit, so that six repeating coils are required. All such circuits required rather good metallic circuits to work properly, but gave lots of circuits with little copper. Phantom circuits were common for telephones as long as pole lines existed.


Tape

Charles Wheatstone developed a telegraph transmitter that used a punched paper tape rather than a live operator. This made rapid sending practical (Thomas Edison made many experiments along this line, using a chemical receiver). This developed into the perforator, a device with a keyboard that punched the holes in the tape, and the transmitter distributor that read the tape and transmitted the characters over the line. This separated the typing and the transmission, with several advantages. One was that a message to be sent to several destinations (weather, news, lottery results) had only to be perforated once, and could then be sent as often as desired with no additional work. Also, several perforators could be working in times of heavy traffic, with the messages ready to send when a line became available.

The paper tape was a buff-colored, sturdy heavy paper. There was a longitudinal row of smaller holes for the sprocket wheel that drove the tape, and horizontal rows of up to five larger holes. A hole represented a marking condition of the line. On transmission, contact fingers would make contact through the holes to a conducting surface. These would be selected one by one in the proper time intervals by a rotating shaft as in the usual keyboard transmitter. The low-order bit was on the left, the high-order on the right.

There was also the reperforator, that punched a tape as the message was received over the line. This tape could then be used to forward the message, using any transmitter distributor. Paper tape reduced errors in relaying, a traditional and fertile source of mistakes, to nearly zero.

Paper tape was used for communication with a digital computer, such as the IBM 1620 (a discrete-transistor machine with roughly the capabilities of an Apple II). One typed at a perforator, then fed the tape into a distributor. It was necessary to make multiple passes with the source tape, and then a reperforator fed you the object tape, which you then submitted. This took place in the early 1960's.

The Universal Asynchronous Receiver and Transmitter

The semiconductor device that performs the major functions of a teletypwriter is called a Universal Asynchronous Receiver and Transmitter, or UART. The keyboard and printer are not included, of course, but when added the system will emulate a teletypewriter. A typical UART, the AY3-1015D, can be set for five data bits and 1-1/2 stop bits by tying pin 36 high, and pins 37-38 low. A bit rate of 50 bits/s is established by applying a clock of 800 Hz. A typical bit rate generator, the AY5-8116, produces 800 Hz as its lowest-frequency output with a crystal of 5.0688 MHz. This gives a bit time of 20 ms, about the same as the 22 ms that was once standard. Any actual teletypewriter can probably be adjusted to receive 50 bits/s with little trouble.

Of course, the proper codes must be sent to the UART, and they are not the usual ASCII codes. My Cherry B70-05AB keyboard produces ASCII codes. An EPROM could be used to make the conversion, using the ASCII as the look-up address. This would allow the addition of the key codes peculiar to the teletypewriter (such as FIGS, LTRS). A computer could also handle the conversion easily. Conversion will also be necessary on the printer end, and can be handled the same way.

References

Department of the Army Technical Manual TM11-665, Fundamentals of Telegraphy (Teletypewriter) (Washington, DC: Dept. of the Army, 1954). A good explanation of the mechanism and circuits. The illustration of the teletypewriter is from page 31, Figure 31.


Return to Telegraph Index

Composed by J. B. Calvert
Created 8 August 2002
Last revised