energy can be inductively transferred to an aerial wire so as to radiate continuous electric waves from it. The arc must be in a certain active condition determined by its length and the strength of the magnetic field in order to produce oscillations. Many investigations have been made to elucidate the working of this oscillation-producing arc.
A fairly complete list of these papers is given by P. O. Peder- sen in the Proceedings of the Institute of Radio Engineers (United States), vol. v, p. 255, Aug. 1917.
In order that the arc may be active, i.e. produce oscillations in the condenser circuit, it must be drawn out to a certain length, and the transverse magnetic field must have a certain optimum value, which depends upon the density of the gas in which it is immersed and on the frequency of the oscillations and the strength of the direct current feeding the arc.
Under best conditions the effective or root mean square value of the oscillatory current is iV2=o-7 of the strength of the direct feeding current. Thus, if the arc is fed with 100 amperes (D.C.), it should give 70 amperes (A.C.) in the oscillation circuit under best conditions: the possibility of this transformation is the result of the negative slope of the characteristic cucve of the direct cur- rent arc, viz. that an increase in arc current is accompanied by a decrease in electric potential difference and vice versa. Also the necessity for maintaining round the arc a non-oxygenic atmosphere, or one consisting of hydrogen or carbon hydrides or oxides, is due to the fact that in these gases the arc characteristic has a steeper downward slope than in air (see W. L. Upson, Phil. Mag. July 1907). The transverse magnetic field is requisite suddenly to extinguish the arc at each oscillation, and so produce an electromotive force in the inductance coil which recharges the condenser in the reverse direction. Broadly speaking then, the operation which takes place is as follows: if the arc is burning steadily and the condenser is shunted across the electrodes, the result is to rob the arc of some current. Hence the potential difference (P.D.) of the arc electrodes increases. This, however, continues the charging of the condenser in the same direction. Then the latter discharges back through the arc and this lowers the P.D. of the electrodes.
The study of the oscillatory arc by means of the oscillograph by H. Th. Simon, H. Barkhausen, A. Blondel, and P. O. Pedersen has shown clearly the nature of the operations taking place. If no magnetic field, or a weak one is employed, and if the arc is in air only, feeble oscillations are set up in the condenser circuit, and the cur- rent through the arc is a pulsatory unidirectional current. This is the case of the Duddell or musical arc which has no use in wire- less telegraphy. If a stronger magnetic field is used and if the arc is in a hydrogen or coal gas atmosphere, then much more power- ful oscillations are produced, and when the R. M. S. value of the condenser current is equal to, or greater than 70% of the direct current the arc current just falls to zero, or is extinguished at each oscillation. The function of the transverse magnetic field is then to blow out the arc by forcing the stream of electrons out- ward, and the effect of this sudden rupture is to create a strong adjuvant or assisting induced electromotive force in the inductance coil in the condenser circuit. This continues the arc current in the same direction, and the condenser thus becomes charged in the opposite direction. The process then repeats itself and we have powerful oscillations produced in the condenser circuit.
Although the condenser current is a sinusoidal current, and the arc current has the same form, yet owing to the shape of the dynamic characteristic curve the potential difference of the arc electrodes is an irregular curve with sharp peaks corresponding to the instants of cessation and recommencement of the arc current.
The practical construction of a Poulsen arc generator involves therefore a large electromagnet having poles which project into a box which can be kept full of alcohol, or kerosene vapour or else coal-gas. Into this box project also two electrodes, one of copper, through which water circulates to keep it cool, and one of hard carbon which is kept in slow rotation by a motor. The poles of the magnet are shaped bluntly conical so as to concentrate a pow- erful magnetic field transversely to the electric arc which springs from the copper (+) to the carbon (-) electrode. The arc is created by a direct current dynamo giving a voltage of 500 or more (see fig. 2). A separate shunt-wound dynamo is often employed to excite the electromagnet. In the circuit of the arc supply dynamo choking coils are inserted, and a circuit comprising a condenser of capacity C and an inductance (L) is connected as a shunt to the arc. If the capacity C isjneasured in farads and the inductance in henrys then the ratio VL/VCis a function of the dimensions of a resistance reckoned in ohms, and should have some value of about 500 ohms or so.
Means must be provided for adjusting the magnetic field to its optimum value (Ho) which depends on the frequency (n) of the oscillations produced, where n is nearly equal to i/2irVLC, or upon the radiated wave length X (in metres) where nX = 3OO mil- lion metres.
P. O. Pedersen states that Ho=o/X-6 where a and b are cer- tain constants (see " The Poulsen Arc and its theory " Proc. Insti-
tute Radio Engineers, United States, vol. v, p. 255, Aug; 1917).
where P is the power fur-
L. F. Fuller states that Ho =
nished to the arc and K is a constant depending on the surrounding gas or vapour. For kerosene K=4-23; for ethylic alcohol K = 8-5. For a power of 50 kilowatts and a wave length of 7,000 metres the arc in alcohol requires a field of 8,300 C.G.S. units and for a wave length of 20,000 metres and a power of 1,000 kilowatts the field must be 13,500 C.G.S. units. Hence as the air gap is large (generally several centimetres) extremely large magnets are required. For the 1,000 kilowatts arc plant the magnets weigh 80 tons and for the 500 kilowatts plant 65 tons. For smaller sizes of plant the magnet is of the open circuit type and for the larger of the closed circuit type. The arc chambec and magnets have to be cooled by water or oil circulation. From 100 kilowatts size and upwards this arc generator is very widely used in long distance radio sta- tions. It labours, however, under the disadvantage that the signalling cannot be conducted by starting and stopping the arc but only by throwing the aerial out of tune or by deflecting the energy into a non-radioactive circuit. Hence power is equally consumed whether message signals are being sent out or only spacing waves.
FIG. 3. Modern thermionic generating valve ; showing the cylin-
drical anode and metal gauze grid, as made by the Marconi-Osram
Valve Company. (By permission of Marconi s Wireless Telegraph
Co., Ltd.)
The Thermionic Valve. The third type of high frequency electric oscillation generator which has become of great impor- tance in the last five years is the thermionic valve, which is i development of the Fleming valve (see 26.537).
The Fleming valve comprises a glass bulb, highly exhausted of its air, and contains a carbon or metal filament which can be rendered incandescent by an electric current. Around the fila- ment is placed a metal cylinder carried on a wire sealed through the bulb. The peculiar property of it is that, when the filament is incandescent, the space between the filament and cylinder will conduct negative electricity from the filament to the cylinder but not in the opposite direction. Hence the name "valve" given to it. It can, therefore, be used to separate out the two constituents of a high frequency alternating current and " rectify " them into a direct current: This valve was ex- tensively used as a detector of electric waves in wireless telegraphy from 1904 onward as described later. In 1907, Lee de Forest in the United States, after he had become acquainted with Fleming's invention of the valve and its use in wireless telegraphy, added to it an additional element in the form of a grid or zig-zag of wire placed between the cylinder and the filament but carried on a separate terminal. He thus made a so-called three-electrode thermionic valve, a name sometimes shortened into triode.
