In 1909, Cyril F. Elwell bought the United States rights to use Poulsen's arc-transmitter, which had been invented in 1903. Elwell became the Chief Engineer of a San Francisco, California firm, the Federal Telegraph Company, set up in 1910 to install Poulsen arc radio stations. At this same time Lee DeForest was also trying, unsuccessfully and without first obtaining the rights from Poulsen, to develop arc-transmitters. And a year after this article appeared, DeForest, with his New York company collapsing, became an employee of Federal Telegraph for a two-year stint. Part of his efforts included working on Poulsen's "rapid transmitter" and "rapid receiver", which tended to work somewhat better in theory than in practice.
Journal of Electricity, Power and Gas, April 2, 1910, pages 293-297:
THE POULSEN SYSTEM OF WIRELESS TELEPHONY AND TELEGRAPHY1
BY C. F. ELWELL.
As soon as the methods of signaling through space first given to the world by Marconi were well understood, scientists throughout the world recognized the shortcomings of both the transmitting and receiving apparatus. Transmission was first effected by means of strongly damped oscillations generated by means of powerful sparks from condensers charged by means of large induction coils, in the primary circuit of which a suitable telegraph key was inserted.
Many improvements have been made in this type of transmitting circuit. Commercial transformers working at a frequency which would give the maximum sensitiveness to the telephone receivers at the receiving station have been substituted for inefficient spark coils. Many attempts have been made to suppress the noise of the sparks, which with the increased use of large amounts of power became distressing to the operator besides betraying the message to unauthorized parties. Apparatus doing this successfully also decreases the efficiency of the apparatus. Keys had to be devised to break the necessarily large primary currents and quite an array of electromagnetic and oil immersed keys are now in use. For heavy power working the speed is limited by such apparatus. Sending condensers have been improved both as to bulk, durability and cost.
Signals were received by means of the Branly coherer on which much time and money were fruitlessly spent. Then came the magnetic, electrolytic and thermo detectors with increased sensitiveness and automatic decohering features. But these detectors have not the well defined resistance which is necessary for accurate resonance tuning effects.
The Danish inventor, Valdemar Poulsen, took up the study of the wireless transmission of signals and recognized the fact that further advance depended on decrease of the damping of the oscillations and increase of sensitiveness of detectors. He determined to follow up the generation of undamped waves as being the line on which more selective telegraphy would be obtained and telephony also be made possible. After a profound study of the "singing arc" following in the footsteps of Elihu Thomson and Duddell, he evolved his present type of arc generator. This generator, with suitable capacity and inductance in shunt to the arc, sets up trains of practically undamped waves of frequencies from two hundred thousand to one million per second according to the values of capacity and inductance in the shunt circuit. Not only this, but he has been able to transform as much as 30 kw. of d. c. to high frequency current in the shunt circuit.
With this generator the solution of the problem of telephoning through space was immediately solved. Applied to telegraphy, it gives improved selectivity of the instruments to an extent never reached by spark methods, permits of duplex working, gives great range with small amounts of power, better results over land, better daylight working and, last but not least, a great increase in speed. For the purposes of telegraphy he had to invent a new type of detector which is now known as the "ticker" and which has been shown to be much more sensitive than any other detector. This detector was necessary in order to render the telegraph signals audible because the alternations take place at a speed much above the limits of audibility.
I will take up the pieces of apparatus, which show the mark of Poulsen's genius, in detail and then give you a short indication of the results obtained with them, together with a short note on the theory of action of the generator and ticker.
In Fig. 1 is shown a Poulsen generator. The arc takes place between a water-cooled copper anode and a revolving carbon cathode. The anode and cathode project through two opposite sides of a water-cooled chamber. The arc takes place in the presence of a powerful magnetic field at right angles to the flow of current. The coils can be plainly seen and the poles project through the other sides of the chamber. The small motor revolves the cathode very slowly and prevents a deposit of carbon taking place and so shortening the arc gap, which is maintained at from 3 to 5 mm. in length. The chamber is equipped with inlet and outlet for supplying the arc with a hydrogen-containing gas. On the generator shown there is a sight feed oil cup in which alcohol is placed. Drops of alcohol on being introduced into the chamber are immediately vaporized and this method of gas supply is in use on ship board. The mechanism seen to the front of the generator is for striking and adjusting the length of the arc by hand. In the chamber there is a yoke which is attracted by the magnetic field when the current is switched on and a small copper tip serves to strike the arc. This automatic arc striking feature was devised for wireless telephoning, so that talking and listening could be carried on with ease.
A large amount of heat is produced in the water-cooled chamber which is removed by means of the cooling water. A certain amount of power is absorbed in the regulating resistance in series with the arc. Of the power which is converted into high frequency oscillations, part is dissipated as heat in the capacity and inductance and part is radiated by the antenna.
A wattmeter may be used to measure this radiated energy by using a direct coupled antenna and measuring the watts at some point in the condenser circuit with and without the antenna. The difference will be the watts radiated.
Fleming has shown that if W represents the energy in ergs radiated per second, when the oscillations are persistent, W = 128 A2, where A is the current read on a hot wire ammeter. Thus a current of 2 amperes would give a radiation of 512 watts, showing that when working with persistent oscillations and open antennae, we can use very small antenna currents, and obtain powerful radiation effects.
The generation of high frequency alternations in a shunt circuit to a continuous current arc is somewhat as follows:
If the arc is steady and is then shunted by a condenser, the current rushes into the condenser and momentarily robs the arc of current, causing the potential difference in the carbons to rise and continue charging the condenser. When the condenser is full the arc current returns to its former value, the potential difference falls, and the condenser discharges from the arc, and the cycle repeats itself. A part of the energy of the continuous current arc is thus changed into the energy of electric alternations in the condenser circuit.
The characteristic curve of the arc is, as is well known, a falling characteristic, i. e., the voltage decreases as the current increases, and for a carbon-arc is comparatively flat. It has besides a persistency which renders it irresponsive to rapid variations of current. Hence only slow alternations can be obtained from a large current carbon-arc.
With the Poulsen arc, with its carbon negative and cooled copper positive, immersed in hydrogen, a very steep characteristic is obtained and one which responds to exceedingly rapid variations of current through it. A condenser of small capacity may be employed in the shunt circuit and yet convey to it a considerable amount of energy because of the large variation of the difference of potential at the arc caused by small arc current variations. So alternations of high frequency can be produced.
This theory is confirmed by the study of small current carbon-carbon and carbon-aluminum arcs in air. For they have steep characteristics and can produce alternations of high frequency. The theory of the part played by the hydrogen on large current carbon and metal arcs is not yet well understood. It appears to be partly due to its greater conductivity compared to air, thus helping to cool the arc electrodes. Poulsen also considers that hydrogen increases the conductivity of the arc.
Practically, the ticker consists of nothing but two fine crossed gold wires, which are vibrated at the rate of 100 vibrations per second, by means of an electromagnet or clockwork. This may be connected to a secondary circuit which is coupled electromagnetically with the primary circuit as in Fig. 2, or in many other ways.
The theory of action is about as follows:
A indicates a receiving antenna or aerial circuit from which alternations are induced in the coil B, which, together with the condensers C and D, constitutes a closed resonant circuit; R may be any form of detector but an ordinary telephone receiver is usually used; I is the interrupter mentioned above and is connected to connect condenser E in parallel with condenser D. When the contact at I is open and assuming that the resonant circuit B C D is tuned to resonance under these circumstances, intense alternations will appear in this resonant circuit B C D, without passing through the telephone receiver R, because of its enormous reactance to high frequency alternations. If now the interrupter I closes the circuit and throws in this condenser E the accumulated energy in the resonant circuit B C D will discharge itself suddenly through the telephone receiver R. The reason for this action is approximately as follows:
While the coil A and B are in resonance the condenser C is charged and discharged at a rate corresponding to the frequency of the alternations. B therefore offers no opposition to this charge and discharge but assists it and maintains the intensity of the alternations. If, however, the condenser has a charge when B is thrown out of resonance with A because of the closing of I and the insertion of more capacity E the discharge of the charge in C will be opposed by B and the charge will have to find another path which it does through R. This discharge takes place in a minute fraction of a second, thus producing a sharp tap in the receiver.
Rapid Telegraph Transmitter.
Fig. 3 shows the rapid telegraph transmitter which is operated by means of a punched tape. The tape has a series of small holes down the center. The holes on each side of this central line are punched by hand and those on one side represent the dots, while those on the other represent the dashes of the Continental code. The central line of holes engages the teeth of a sprocket wheel which serves to feed the tape forward at a regular rate. The tape gear wheel has a number of very small and light radial pins, which tend to fly out except when they are held in position by the tape. Wherever a hole occurs a pin is allowed to spring outwards. These pins are mechanically connected to larger pins on a further attachment of the spindle which fly out when the smaller pins are actuated by the tape. Spring contacts are in series with a set of brushes which press on the segments of three rotating commutators, one of which has a comparatively large number of alternate conducting and insulating segments, and is reserved for the dots, while the other two have longer spacings of commutator segments, which are kept for the dashes. In this way all the actual making and breaking of the current is carried on on these larger segments, while the tape controls the whole apparatus by means of the lighest possible form of mechanical construction. This reduces the effect of inertia to the lowest limit. There are 72 pins, each representing a dot and a space. An average word has five letters, so it is possible to transmit three words for one turn of the transmitting combination and the speed of the machine, which is driven by a direct current motor, can be varied between the limits of ordinary hand speed to a transmission of 300 words or more per minute. The practical limit at present being in the receiver and not in the sender.
Rapid Telegraph Receiver.
A complete rapid receiver is shown in Fig. 4. It consists essentially of an Einthoven "string" galvanometer in which a gold string is used in connection with a thermocouple. The absence of inertia permits the string to follow the rapid impulses sent out by the rapid sender. A coating of soot is placed on the wire and the wire itself is mounted in the beam of a Nernst or arc lamp. A suitable optical condenser throws the light on a narrow slit behind which moves a band of photographically sensitized paper. The shadow of a small portion of the wire as it vibrates to and fro in response to the signals from the sending station is thus imprinted on the band, which is then drawn, first through a developing bath, then through a fixing bath, and then through water to wash it. The message may be read on the developed band as soon as it emerges from the light tight box and may be kept as a permanent record. The signals are read above the zero line which is traced by the shadow of the wire when no impulse is present. A short impulse makes a dot and a long impulse a dash.
The problem of wireless telephony involves essentially three things:
1. The production of undamped or persistent waves in a transmitting antenna.
2. Means for modulating these waves in accordance with the wave form of the spoken voice.
3. Means for detecting the waves at the receiving end and their reproduction into articulate speech.
The Poulsen generator offered a means of supplying the undamped waves in the transmitting antenna and it was only necessary to connect a microphone at or near a node of current in the antenna to supply a means of modulating these waves in accordance with the wave form of human speech.
At the receiving end almost any self-decohering detector will do, but the production of good clear articulation depends quite a little on the degree of coupling of the primary circuit with the secondary circuit. This also applies to the sending circuits in which quite loose coupling is employed.
Poulsen has transmitted good, clear, articulate speech over the 180 miles between Esbjerg and Lyngby, Denmark. Majorana claims to have done 312 miles over water with a specially constructed microphone of his own devising. More recently I have carried on successfully two way working between Stockton and Sacramento, California, a distance of 50 miles over land, and while working between these two stations was heard by St. Helena and Palo Alto, distances of 75 and 85 miles respectively.
There is no doubt that wireless working gives telephony of a higher grade than wire working. There is absolutely no noise in the receiver until spoken words are heard. To one who has talked over long distance wire lines with considerable induction this feature readily appeals. Low resistance receivers are used and expensive high resistance receivers are not necessary.
With the Poulsen generator of continuous waves it is possible to telegraph at hand speed, i. e., 25 words per minute in many ways. For example, it is possible to signal by:
(a) Short circuiting a resistance in the generator circuit.
(b) Short circuiting a resistance in the antenna circuit.
(c) Making and breaking the arc.
(d) Altering the length of the arc.
(e) Altering the strength of the transverse magnetic field.
(f) Altering the flow of gas through the arc.
In practice Poulsen short circuits a turn or two of the sending inductance by means of an ordinary Morse sending key. The absence of the spark permits of the use of an ordinary key when telegraphing 2000 miles, for the current is even then quite small.
For receiving, Poulsen uses the ticker which has the great advantage of not being receptive to ordinary damped wave signals.
The tuning possible with the Poulsen arrangement for telegraphy is extremely close. One-half to one per cent change in the capacity of the resonant circuit is readily noticed on the received signals. Duplex working has been carried out with 3.9 per cent change in wave length.
The rapid wireless telegraph transmitter and receiver have already been described. In practice the transmitter is connected in, just where the Morse key would be for hand speed telegraphy. Good, clear, readable records have been received over 180 miles, mostly land, at the rate of 300 words per minute. Over 600 miles good records have been received up to 150 words per minute. As a means for handling large quantities of business and with a record at both transmitting and receiving stations the rapid system has a good future. Poulsen estimates that he can handle 100 words per minute across the Atlantic with a 60 kw. generator and suitable antenna.
Advantages and Future Possibilities.
In the first place the absence of all noise is brought home forcibly in a Poulsen station. It seems hard to believe that anything is being done at all. The key may be of the ordinary Morse type, for the currents handled are quite small even for large distances. There are no insulation difficulties, for the voltage at the top of the antenna is not estimated to be over 3000 volts. The sending helix may be handled quite without shock, even though the voltage be over 1000.
Very small capacities are used with heavy power working, eliminating a source of expense and a very bulky part of large "spark" stations. The capacity in connection with a 12 kw. set is about .0017 microfarad. At the Cullercoats, England, station the condenser takes up less than a tenth of the space occupied by the condensers for a "spark" system of the same power installed in the same stations.
Undamped waves of small amplitude are less obstructed by atmospheric conditions and suffer less absorption over land than damped wave trains. For example, in coming around the north of Scotland the undamped wave signals are picked up long before the damped wave signals of equal power.
The form of the resonance curve of the receiver circuit depends on the decrement of the transmitter and receiver. If the transmitter is undamped, a very small change in the period of the receiver will put it out of tune, hence, a receiver circuit can be employed which is sensitive to undamped waves of some exact period, but which is exceedingly nonresponsive to waves differing by a very small fraction of one per cent in wave length from the syntonic value.
In the matter of energy, good signals have been transmitted over 500 miles with 1 kw., and over 2000 miles with 6 kw., and a limited antenna. The efforts of Marconi and others to reduce the damping of their wave trains is evidence that undamped wave trains will be the means of communication of the future.
In telephony at the present time better articulation is obtainable than with wires and it is quite probable that a method of obtaining secrecy will soon be devised.
In telegraphy there is no doubt that the Poulsen system has great range for little power and that it works readily over land and in daylight. I look to see the present records of distance now held by spark methods broken by stations using the continuous waves. The present speed of the rapid telegraph sender is dependent on the receiver, but a newer type of rapid telegraph detector is coming out which will no doubt result in the handling of greater speed than 300 words per minute. Great advances can be expected in the next five years in wireless working, but they will be along the lines of work with continuous wave trains
1Paper read before San Francisco Section A. I. E. E. March 25, 1910.