It was recognized by the early workers in wireless telegraphy and telephony that if it should be found practicable to modify or modu late the high frequency ether waves utilized in wireless telegraph operation, it would be pos sible to transmit speech without wires to a dis tance approximating that of wireless telegraphy. A difficulty, however, in availing of high fre quency electromagnetic waves for this purpose has been that as ordinarily obtained from the discharges of a condenser at the spark gap in wireless telegraphy such waves are intermit tent and their amplitude is not uniform, but is quickly damped,,as noted in the article Wurx Less TELEGRAPHY. Thus, in case of an alter nating current generator developing a current of, say, 60 cycles, or 120 alterations per sec ond, there will be 120 sparks per second at the gap if at each alternation the oscillation circuit is charged to sparking potential, which is not always the case. Assume that each dis charge of the oscillation circuits gives rise to oscillations of a frequency of one million per second in that circuit, with a consequent wave length of 300 meters. The exact number of os cillations and wave length depends on the in-. ductance and capacity of the oscillation circuit. (To conform to the word inductance the capac ity of a circuit is now frequently termed capac itance). As the capacitance of the aerial is usually supplied by the aerial wires and is not readily variable, the frequency rate is varied by placing an adjustable inductance in the aerial circuit in the shape of a spiral coil of wire more or less of which may be inserted in or removed from the circuit. The capacity of the aerial is sometimes reduced, for •short wave signaling, by placing a condenser (capacity) in the aerial circuit. In electrical diagrams a variable or adjust able condenser or inductance is conventionally indicated by an arrow through the apparatus. In the case in point if the oscillations were maintained throughout the interval between dis charges there would be 8,334 oscillations per spark, that is, the quotient of 1,000,000 divided by 120. Since, however, the oscillations in such a circuit die out very rapidly, due to the damp ing effects mentioned in the article WIRELESS TELEGRAPHY, there are only two or three strong oscillations for each discharge in highly damped circuits and 10, 20 or 30 strong oscillations in less strongly damped circuits. Hence oscilla tions are only maintained during a brief inter val between spark discharges, leaving approxi mately the one hundred and twentieth of a second between such discharges during which there are no oscillations in the oscillation circuit and at that time no waves are radiated from the aerial. In wireless telegraphy these inter. mittent or group oscillations after reception by a detector are observable in the telephone head receiver as a continuous tone or buzz which is broken into dots and dashes of the Morse alpha bet by a telegraph key. If the attempt were made to superpose vibrations corresponding to voice waves upon these intermittent or broken oscillations, it is evident that portions of the voice waves set up by the transmitter would be lost.
Kennelly has mathematically demonstrated that speech might be transmitted telephonically with harmonics, or overtones, not exceeding 2,000 per second, but Fessenden has found by experiment that for satisfactory speech a fre quency of at least 20,000 per second must be employed, otherwise disagreeable noises will be heard in the telephone receiver. Therefore, continuous or sustained oscillations of a fairly high frequency and uniform amplitude are es sential upon which to superpose the voice fre quencies for successful wireless telephony. To this end fairly successful efforts have been made by different inventors to produce a machine generator, or alternator, capable of delivering an alternating current of very high frequency notably by Fessenden, Goldsmith and Alexan derson. Owing to the mechanical difficulties of
construction of such machines it is perhaps a question if they will prove commercially suc cessful in general use. Nevertheless such ma chines have given and are giving good service in a number of instances. An idea of the dif ficulties of construction of high frequency alter nators will be obtained when it is pointed out that to obtain 100,000 oscillations per second in a 600-pole rotor having a diameter of two feet, a speed of 20,000 revolutions per minute is necessary. Fairly full descriptions of high frequency machine generators will be found in books mentioned in Bibliography. Consult paper by Alexanderson, Proceedings American Institute Electrical Engineers (October 1919).
Another method of obtaining sustained elec trical oscillations is known as the singing arc, due to Duddell, who discovered that when an arc lamp is in shunt with a capacity and an in ductance of given proportions, a musical tone is set up in the arc. Ultimately Duddell ob tained frequencies of 40,000 per second from this source. An explanation of this phenomenon is that when the shunt circuit is completed a cur. rent flows from the arc circuit into the conden ser or capacity circuit which decreases the cur rent flowing in the arc. This increases the electromotive force between the terminals of the arc, causing still more current to flow in the condenser circuit raising its electromotive force above the normal voltage of the arc. Con sequently the condenser begins to discharge back into the arc, increasing the current therein and reducing the electromotive force between its terminals, when the reverse process is set up, and in this way continuous oscillations are maintained in the arc and shunt circuits.
When sustained oscillations are compara tively uniform in amplitude and their frequency is above audibility these oscillations are not heard in the telephone receiver. But, while the telephone receiver or at least the human ear will not respond to such high frequencies, if the amplitude or contour of the oscillation waves be modulated by speaking into a micro phonic transmitter placed in the oscillation cir cuit, or in the aerial circuit of a wireless sys tem, a telephone receiver will under proper con ditions reproduce the speech spoken at the transmitter, virtually as speech is reproduced by modulating the amplitude of a beam of light, or a telegraph current in the case previously cited. The terms °radio frequency° and •audio frequency" are now commonly used to indicate the inaudible high frequency oscillations and the audible or low frequency oscillations, respec tively, in wireless telephony.
The discovery of the oscillating arc paved the way for the use of sustained oscillations of high frequency and by means of modifications of Duddell's device much progress has been made in wireless telegraphy and wireless tele phony. Such devices are now sometimes termed arc radio transmitters or arc converters.
An important improvement in the Duddell oscillating arc is that due to Poulsen. In the Poulsen arrangement the positive electrode of the arc is a water-cooled copper cylinder. The negative electrode is carbon. The arc is placed in a strong magnetic field which tends to blow out the arc, after the Elihu Thomson effect. The source of current supply for the arc is a direct current generator supplying from five kilowatts to 100 kilowatts at 500 to 600 volts. The arc converts the energy of the direct cur rent machine into alternating currents of a fre quency of say 60,000 per second. The practical efficiency of the arc converter is 33 per cent to 50 per cent of the input. For further inter esting details of this arc converter and its util ity in Wireless Telegraphy, consult an article in Electrical World (30 Aug. 1919, p. 452).