From Current Literature.

The thermionic tube as a physical instrument.

The properties inherent in the thermionic tube (otherwise—the amplifying tube, the “valve,” the triode), namely its amplifying and detecting capacity, its rectifying action and, finally, its capacity for generating undamped oscillations, make this tube an exceedingly valuable physical instrument. In the present article we give a brief (and not altogether exhaustive) survey of physical measurement methods based on the use of such a tube, and of physical investigations carried out with its aid. Details concerning the tube itself, as well as concerning amplifiers and generators, may be found in ¹), ²), and ³). In what follows we have in mind almost exclusively the small tube used for receiving radio signals.

The most commonly used arrangement of the three essential parts of the tube (the anode, the hot cathode, and the “grid,” i.e. a mesh or lattice electrode whose potential relative to the cathode regulates the thermionic flow from cathode to anode) is such that the heated straight “filament” (cathode) is surrounded by a cylindrical “grid,” and the latter by a cylindrical anode (“cylinder”). The relation between the filament-grid potential and the anode current, with unchanged voltages on the anode and at the ends of the heated filament, is quite definite and single-valued (the “characteristic of the tube”); therefore, if the characteristic is known, it is possible, directly from the readings of a milliammeter in the anode circuit, to determine (a slowly varying) potential difference between filament and grid: this is the so-called “thermionic electrometer” of Holague ¹); in particular, the potential difference measured may be produced at the ends of an ohmic resistance, in which case an amplification of action by a factor of 5–10 is obtained. By inserting a suitable resistance into the anode circuit and leading its ends to the filament and grid of a second tube, etc., we obtain multiple “cascade” amplification. By a similar method Pike ⁴) measured the photoelectric effect, achieving an amplification of 5000 times. Among us this method has been used by Termen ⁵), as well as by Rzhevkin and Vvedenskii ⁶) and Mints ⁷).

If, in such an arrangement, the milliammeter is replaced by a telephone, connecting the grids of two, three, etc. tubes not directly, but through a small capacitor [Arni-[[unclear: name]] ⁸)], and the grid potential is constantly returned to the filament potential (this is achieved by connecting the grid with the filament through a large resistance, which does not hinder instantaneous increases of the grid potential), we obtain a high-frequency amplifier, with one of the tubes acting as a detector [detector connection ⁸), ⁷), ²), and ³)]; at the present time the resistances in the anode circuit are often made inductive. For low-frequency amplification it is more advantageous to increase the potential oscillations transmitted to the grids of the tubes by means of transformers ¹), ²), ³). The attainable amplifications are of the order of 1000. With such an amplifier Barkhausen and Tucek ⁹), by connecting to the grid and filament a coil with slowly remagnetized iron, detected the discontinuity of the magnetization process, manifested in the appearance in the telephone of a characteristic rustl-

...or crackling. The phenomenon is easily detected with an ordinary amplifier. Schottky^40) showed that, with sufficient amplification, one can detect the natural oscillations of a certain Thomson circuit, connected in parallel with the anode and cathode (hot) of a certain vacuum tube, and interpreted the phenomenon as the action of the “shot effect” (Schroteffekt), i.e., the “fluctuation” (Schwankung) in the number of electrons emitted from the filament per second. Schottky’s theory shows that, from the magnitude of the amplitude of oscillation of the circuit, it is possible to determine the charge of the electron; however, Hartmann’s^41) experiments did not confirm this.

The simplest generator circuit is as follows: in the anode circuit, besides the battery, a self-inductance \(L\) is inserted, and in parallel with it a capacitance \(C\) (an oscillatory circuit); in the grid circuit there is a second self-inductance, coupled with \(L\) in such a way that an increasing current in \(L\) induces a \(+\) potential on the grid and thereby increases the current in the anode and \(L\); then the increasing current in \(L\) will lock the anode current. In general, the number of generator circuits is very large^4). Usually the period of oscillation at small damping is determined by the oscillatory circuit, i.e. by the formula \(T=2\pi\sqrt{L(C+\delta)}\), where \(\delta\) is a certain additional capacitance^4); on the determination of \(C+\delta\) see, for example, ^25). Depending on \(L\) and \(C\), the period \(T\) may vary from \(0.2\) sec. [White^42)] and lower down to approximately \(5\cdot 10^{-8}\) sec. \((\lambda \approx 15\,m)\); under exceptional conditions a wavelength up to \(6\,m\) is reached [White^42)] and even up to \(1.1\,m\)^43).

With large \(L\) and \(C\) we obtain an audio generator—an extremely convenient source of sound of constant intensity and pitch; moreover, both the intensity and the pitch are easy to vary at will to any degree.

As a source of high-frequency oscillations, the tube generator is irreplaceable, since, owing to the constancy of its action and the simplicity of handling it, it makes it possible to transfer to high frequency the methods and skills developed in working with ordinary alternating current. For example, Lertes^44) obtained a high-frequency rotating electric field, by means of which he attempted to measure the moment of Debye dipoles. By weakly coupling the resonant circuit \(A\) with the generator, Vvedenskii and Teodorchik^45) measured the permeability of iron \(\mu\) at high frequencies directly from the change in the resonance capacitance when a bundle of iron wires was introduced into the self-inductance of circuit \(A\); the same authors succeeded in determining \(\mu\) also from the heating of the wires (the Klemenčič method for damped oscillations) and from the ponderomotive action of an alternating field on a bundle of iron wires suspended on a quartz fiber. Teodorchik^46), from the change in resonance capacitance upon connecting a capacitor with a liquid dielectric, measured a number of dielectric constants of alcohols and even of distilled water without any compensation of conductivity; this latter circumstance makes this method more convenient than the well-known Nernst bridge method.

In radio engineering, reception of undamped (and, consequently, inaudible in a telephone) oscillations is widely employed by the “beat method,” in which the signal is made audible by producing rapidly following and sound-producing beats between the oscillations of the signal and the local undamped oscillations. A slight detuning of one of the circuits that gives the beat tone (for example, bringing the hand closer) causes a very noticeable change in the tone. The latter can be measured by arranging a “secondary interference” of the beat tone with some tone of constant pitch (a tuning fork or an audio generator) and determining the change \(\Delta N\) in the number of “secondary” beats (i.e., between two acoustic oscillations) when \(L\) and \(C\) of one of the high-frequency circuits of frequency \(\nu\) are changed. Then

\[ |\Delta N|=\nu \Delta L/2L=\nu \Delta C/2C, \]

for example, at \(\nu=\frac{1}{T}=2\cdot 10^{8}\ \mathrm{sec}^{-1}\) we obtain \(\Delta N=1\) already for \(\Delta L\) or \(\Delta C=10^{-8}L\) or \(C\). By this method Whiddington^47) measured changes in the distance between the plates of a capacitor of the order of \(10^{-8}\) cm; Herweg^48), the dependence of the dielectric constant on the field strength (Debye dipoles). This same method was developed by Pungs and Preuner^49) and Hammer^50), while Vvedenskii and Teodorchik^21), ^15) determined \(\mu\) of iron, nickel, and steel wires at wavelengths from \(54\,m\) to \(705\,m\), measuring \(\Delta L\) from the introduction into the self-inductance coil of 4–6 thin wires.

The enormous sensitivity of the method to any change in capacitance (the approach or even simply the motion of the observer’s body, etc.) makes it necessary to arrange electrostatic shielding or to control the apparatus at a distance, by means of glass rods; on this sensitivity is based the qualitative application of the method to musical-instrumental purposes, made by Theremin.

For comparatively considerable changes in \(L\) or \(C\), a simplification of the method is possible: namely, the two interfering generators are first tuned to unison (then the sound of the beats disappears). Then one connects, for example, the capacitance to be measured in parallel with the capacitance \(C\) of one of the circuits and again restores resonance, changing \(C\) by \(\Delta C\). This makes it possible, for example, accurately to graduate small capacitors or to measure dielectr. const. \(^{30}\). Jackson \(^{31}\), measuring in this way the dielectr. const. of complex ethers, in particular upon changes in their state of aggregation.

The method of Falkenberg \(^{22}\) is based on another principle: two coils of identical circuits I and II are coupled with the generator, in one of which the capacitance \(C\) whose variation is measured is placed; I and II act on a new circuit \(B\), the capacitor of which is connected with the grid and filament of an amplifier. At resonance of all circuits, owing to the chosen coupling, I and II compensate each other; in the circuit of the last tube a resistance \(R\) is placed, and in parallel with it a galvanometer, in which, at resonance of I, II, and \(B\), there is no current (owing to a special compensating potentiometer). When the resonance is detuned (by a change in \(C\)), the current in \(R\) changes, and the galvanometer is deflected, which makes it possible to determine \(\Delta C\). The sensitivity of the method is about \(2 \cdot 10^{-8}\), but it is very capricious, little suited to high (i.e. \(\lambda < 1000\,m\)) frequencies and, apparently, has found no application.

In generation, a weak constant (properly, pulsating) current flows to the grid of the tube. When a resonant circuit coupled with the generator is tuned to resonance with the generator, this current decreases. Theremin \(^{6}\) used this phenomenon also for measuring small \(\Delta C\); in particular, on this principle he constructed an instrument for determining the strength of weak sounds (a condenser microphone).

If in the ordinary generator circuit, immediately before the grid, one places a capacitor \(C\), shunted by a large resistance \(R\), then continuous series of oscillations break up into separate trains, separated by more or less prolonged pauses (Armstrong \(^{8}\), Rzhevkin \(^{23}\)). Beatty and Gilmour \(^{24}\), using pauses so small that the interruptions produced in a telephone connected with the generator a continuous tone, obtained a definite, but rather complicated dependence of the pitch of the tone on \(R\) and \(C\). Rzhevkin and Vvedensky \(^{25}\) explained the phenomenon by the accumulation on the grid of a negative charge which “locks” the oscillations; these arise only as the charge leaks through the large resistance \(R\) according to the ordinary law

\[ q = q_0 e^{-\frac{t}{RC}}; \]

this was, incidentally, checked by a direct measurement of \(Q\) by the method of the “thermoelectron electrometer.” The durations of the pauses in these experiments proved to be almost strictly proportional to \(R\) and \(C\) \((7 \cdot 10^8 \le R \le 10^{11}\ \omega)\). Mints \(^{7}\) developed the method of the “thermoelectron electrometer” for determining \(R\) and \(C\).

The tube generator gives a series of overtones \(^{26}\); the latter are usually so weak \(^{1}\) that they can be detected only by an amplifier. By the beat method they can be detected in quantities of several dozen. Vvedensky and Teodorchik \(^{21}\) applied the fact of their existence to determine wavelengths \(<100\,m\) (which are not available on ordinary wavemeters).

Whiddington \(^{27}\) discovered in gas tubes their “proper,” i.e. not determined by the external \(L\) and \(C\), oscillations, and gave the following explanation for this: some small part of the surface of the filament suddenly heats up (for example, by the impact of \(+\) ions) and emits a cloud of electrons, which, rushing toward the anode, forms a cloud of \(+\) ions. The latter, “falling” upon the anode with uniform acceleration, again sharply heat part of its surface. As a result, pulsations of the current arise in the anode circuit \(n\) times per se-

whence, moreover, \(n\) is proportional to \(\sqrt{\dfrac{c}{m}}\) of the ions of the given gas and inversely proportional to the distance filament–grid. Barkhausen and Kurz \(^{28}\) obtained natural oscillations in vacuum tubes with a dense and fine grid at a large (up to 500 V) positive potential on it and at a smaller negative potential on the anode. They regard as the cause of the oscillations the “mechanical” oscillations of a cloud of electrons from the filament to the negative anode and back, about the positive grid; the smallest waves obtained had a length of 43 cm. The phenomenon is not entirely clear even to the authors. Davydovsky, with his tube having two anodes and two grids, obtained waves of the order of 20 cm, superposed on longer ones (overtones?); the principle of action here is apparently different from that in \(^{28}\) and \(^{27}\).

With a negative potential on the grid (\(-2\) to \(-3\) V), current can pass to it only in the presence of \(+\) ions of the gas in the tube; therefore the strength of this current, proportional to the number of ions, and hence to the number of gas atoms, can measure the pressure in the tube [(Schottky \(^{30}\), Möller \(^{31}\)]. Kaufmann and Serowy \(^{33}\) developed a quantitative method (chiefly on the basis of Meyer’s \(^{32}\) investigation of the formation of secondary electrons in \(H_2\) and \(N_2\)). For complete definiteness it is necessary to know the chemical nature of the gas in the tube, which, apparently, can be established, for example, by the electron-impact method of Franck and Hertz. By Kaufmann’s method one can measure small pressures of \(10^{-7}\)–\(10^{-8}\) mm Hg in any vessel connected with the tube.

June, 1922.
B. Vvedensky.

\(^{1}\) H. Möller, Elektronenröhren, Vieweg 1920.
\(^{2}\) Fleming. Thermionic valve. 1919.
\(^{3}\) Articles by Prof. V. K. Lebedinsky and others in “Tel. i telef. b. prov.” and “Radiotekhnik.”
\(^{4}\) C. E. Pike, Phys. Rev., 13, p. 102, 1919.
\(^{5}\) L. Termen, “Radiotekhnik” No. 13, p. 314, 1920.
\(^{6}\) C. Rzhevkin and B. Vvedensky, “Tel. i telef. b. prov.” No. 11, 1921.
\(^{7}\) A. Minz. Reports at RORI 1922.
\(^{8}\) E. Armstrong. Proc. of Inst. Radioeng. 3, No. 3, 1915.
\(^{9}\) H. Barkhausen, Phys. ZS, Sept. 1919.
\(^{10}\) W. Schottky. Ann. d. Phys. 57, p. 541, 1918.
\(^{11}\) C. Hartmann. Ann. d. Phys. 65, p. 51, 1921.
\(^{12}\) W. C. White. Gen. El. Review, Sept. 1916.
\(^{13}\) See note in “Radiotekhnik” No. 14, p. 512.
\(^{14}\) P. Lertes, ZS. f. Phys. 4, p. 315, 1921.
\(^{15}\) B. Wwedensky and K. Theodortshik. Ann. d. Phys. 68, p. 463, 1922.
\(^{16}\) K. Theodortshik, Phys. ZS, 23, p. 344, 1922.
\(^{17}\) R. Whiddington, Phil. Mag. 30, p. 634, 1920.
\(^{18}\) J. Herweg. Phys. ZS, 21, p. 572, 1920; Zs. f. Phys. 3, p. 36, 1920.
\(^{19}\) L. Pungs and G. Preuner. Phys. ZS. 20, p. 543, 1919.
\(^{20}\) W. Hammer. Ber. d. Nat. forsch. Ges. Freiburg, Bd. 22, 1920.
\(^{21}\) B. Vvedensky and K. Teodorchik. “Tel. i telef. b. prov.” No. 13, 1922.
\(^{22}\) G. Falkenberg. Ann. d. Phys. 61, p. 167, 1920.
\(^{23}\) C. Rzhevkin. “Radiotekhnik,” No. 8, 1919.
\(^{24}\) R. Beatty and A. Gilmour, Phil. Mag., Oct., 1920.
\(^{25}\) C. Rzhevkin and B. Vvedensky, “Tel. i telef. b. prov.” No. 11, 1921; Phys. ZS. 23, p. 150, 1922.
\(^{26}\) See, for example, H. Salinger, Ph. ZS. 20, p. 448, 1919.
\(^{27}\) R. Whiddington, Cambridge Univ. Rep., May, 1919; Radio Review, Nov. 1919.
\(^{28}\) H. Barkhausen u. K. Kurz. Phys. ZS 21, p. 1, 1920.

29) A. Danilevsky. Report at the VIII Electrotechnical Congress, 1921.
30) W. Schottky, Arch. f. Elektrot. 8, p. 1, 1919.
31) H. Möller, Arch. f. Elektrot. 8, p. 48, 1919.
32) Meyer. Ann. d. Phys. 45, p. 1, 1914.
33) F. Kaufmann and Fr. Serovy. ZS. f. Phys. 5, p. 319, 1921.
34) L. C. Jackson. Phil. Mag. 43, p. 481, 1922.