Electrical World, December 12, 1914, pages 1149-1152:
 
Operating  Features  of  the  Audion
Explanation  of  its  action  as  an  amplifier,
as  a  detector  of  high-frequency  oscillations
and  as  a  "valve."       By  E.  H.  Armstrong
ALTHOUGH the audion has been in use for several years as an amplifier and a detector of high-frequency oscillations, the explanations advanced to account for its action do not appear to be satisfactory. With the idea of pointing out some features of operation which heretofore do not seem to have been appreciated, the following explanation and oscillograms are given. Figure 1
    The audion is essentially an electron relay; that is, the exhaustion is carried to such a point that the amount of gas present is exceedingly small, and the current between the hot and cold electrodes is entirely thermionic, the absence of gas making impossible the presence of positive ions. The operating characteristic of such a relay is as shown in Fig. 1. This characteristic was obtained in the manner indicated in Fig. 2. Figure 2
    The potential of the grid with respect to the filament was varied in steps between - 10 and + 10 volts, by means of the potentiometer P, corresponding readings of grid voltage and wing current being taken in order to plot the curve of Fig. 1. The characteristic shows that, starting with the grid and filament at zero potential difference, a negative charge imparted to the grid produces a decrease in the wing current and a positive charge imparted to the grid produces an increase in the wing current. This is the fundamental action of the audion when used either as an amplifier or a detector. The reason for this action will appear upon examination of the behavior of an audion of the type shown in Fig. 3. Figure 3
    The wings of the audion were placed symmetrically with respect to the filament, but only one grid was employed. It was found that, under similar conditions of filament temperature and voltage of the battery B2, a considerably smaller current was obtained between the filament and plate on the side in which the grid was inserted. In both measurements the grid was left entirely free of any connection with the rest of the apparatus. Obviously the grid obstructed the flow of the thermionic current. Investigation showed that this was due to the charge accumulating on the grid when exposed to bombardment by the electrons passing from the filament to the wing. The electrons pass readily enough into the grid but cannot easily escape from it, and as a consequence of this negative electricity piles up on the grid. The potential assumed by the grid when exposed to this bombardment may be several volts negative with respect to the negative terminal of the filament, it may be the same as the negative terminal, or it may be positive with respect to the negative terminal, but it will always be negative with respect to the potential of the field in the plane of the grid which would exist if the grid were removed from the bulb. The negative charge on the grid, therefore, impedes the flow of electrons from filament to plate, causing the decrease in the wing current. The placing of a positive charge on the grid from an external source tends to neutralize the negative charge on the grid, thereby permitting an increase in the wing current. The addition of a negative charge to the grid increases the deflection of the electrons and produces a further decrease in the wing current. Figure 4
    An alternating emf impressed between the grid and the filament causes variations in the wing current in the manner indicated in Fig. 4, the positive alternation producing an increase and the negative alternation a decrease in the wing current. This is the action involved in the audion when it is used as an amplifier. Figure 5
    To substantiate the above and other actions, the writer, working in conjunction with Prof. J. H. Morecroft, of Columbia University, has secured oscillograms which substantiate the idea just presented. Fig. 5 shows the arrangements with which the test was carried out.
    The potentiometer P was used to adjust the grid to a potential corresponding to point P at the center of the operating part of the curve shown in Fig. 1. The audion is capable of handling the greatest amount of energy as an amplifier when the grid potential is adjusted to this point.
    Fig. 6 shows the oscillogram of the action as an amplifier. The result bears out the explanation already given. Figure 6
    The action of the audion as a detector of high-frequency oscillations is quite different from its action as an amplifier. Since the incoming oscillations are of too high a frequency to affect directly the telephone receiver, the audion must be so connected and adjusted that the cumulative effect of a group of oscillations in the grid circuit is translated into a single low-frequency pulse or variation in the telephone current. This may be done in two ways, one depending on the non-linear form of the operating characteristic of the audion and the other depending on the so-called "valve" action between hot and cold electrodes at low pressures. Figure 7
    Fig. 7 shows the connection used for operating in the first-named manner. The potentiometer P is employed for the purpose of adjusting the potential of the grid to point M on the characteristic curve of Fig. 1. The action is much the same as in one of Professor Fleming's methods of using his valve. A group of high-frequency oscillations impressed on the grid causes corresponding high-frequency variations in the continuous current in the wing circuit, but owing to the fixing of the grid potential at the lower bend in the curve by adjustment of the potentiometer in the grid circuit, the amplitude of the positive part of the high-frequency current in the wing circuit exceeds the amplitude of the negative part. As the positive half-waves are greater than the negative half-waves, more electricity flows in one direction than the other, and the condenser C, through which the high-frequency current in the wing circuit flows, becomes charged, the side connected to the battery B2 having the positive charge. This charge accumulates in C1 in a relatively short time, approximately that of the duration of a wave train. C1 then discharges through the telephones T, the rate of this discharge being determined by the constants of the telephones and the condenser. It is probable that this discharge is aperiodic or nearly so. In any case the main part of the discharge through the telephones is in the same direction as the current due to the battery B2 and constitutes an increase in the current in the telephones. As this action is repeated for each group of oscillations, a series of wave trains causes what might be regarded (in its action on the telephones) as an alternating current in the telephones superposed on the continuous current and having a fundamental frequency equal to the number of wave trains per second. The action is shown diagrammatically in Fig. 8.
Figures 8 & 9
    If the potential of the grid is adjusted to the upper bend in the curve of Fig. 1, as at point N, the fundamental action will be the same, but the effect of high-frequency oscillations in the grid circuit on the wing current will be reversed. The amplitude of the negative part of the high-frequency oscillations in the wing circuit will exceed the amplitude of the positive part and the condenser C1 will become charged, but in the opposite sense, the side connected to the battery B2 becoming negative. The discharge of the condenser through the telephones will therefore be in the opposite direction to the flow of the continuous current of the wing circuit and will constitute a decrease in the telephone current. Diagrammatically the action is as indicated in Fig. 9. Figure 10
    Oscillograms bearing on these actions were obtained in the manner indicated in Fig. 10. Oscillations were set up by the discharge of the condenser C' through the inductance L', which was coupled with the inductance L of the tuned grid circuit. To permit the use of an ordinary General Electric oscillograph, an oscillation frequency of about fifty cycles per second and a group frequency of two or three cycles were employed. The action of the audion is the same regardless of frequency, provided that the circuit constants are suitably modified to fit the frequency employed. In this case the oscillation frequency of the circuit C'L' fifty cycles and the circuit LC was accordingly tuned to the same frequency. The capacity of C1 was selected to correspond to the low frequency employed. Figs. 11 and 12 show oscillograms taken as indicated in Fig. 10, with the grid potential adjusted respectively to the lower and upper bends of the operating characteristic.
    It will be observed that the telephone current reaches in Fig. 11 its maximum value, and in Fig. 12 its minimum value, when the oscillating current has almost died away. This effect would be shown more plainly with a higher oscillation frequency, but even at the frequency used it is quite evident.
Figures 11 & 12

    To make use of the "valve" action between hot and cold electrodes for the detection of high-frequency oscillations a connection as shown in Fig. 13 is used.
Figures 13 & 14
Figure 15     In this case a condenser C2 is inserted somewhere in the circuit between the grid and filament to prevent the flow of a continuous current between them, and the grid is therefore left free to assume a potential determined by its position with respect to the filament and wing. Usually this will be somewhere near the center of the operating part of the curve of Fig. 1; that is, near point P. Now the action for incoming oscillations, as far as the closed oscillating circuit, filament, grid and condenser C2 are concerned, is identical with the rectifying action of the Fleming valve. An incoming wave train sets up oscillations in the closed circuit LC which are rectified by the "valve" action of the filament and grid, and the rectified current is used to charge the condenser C2. Electrons pass readily enough into the grid but cannot easily escape therefrom, and a negative charge is built up on the side of the condenser connected to the grid. The negative charge thus imparted to the grid cuts down the flow of electrons from the filament to the wing, producing a decrease in the wing and telephone currents. At the end of a wave train the charge in C2 gradually leaks off and the wing current returns to its normal value. The charge and discharge of this condenser take place in the manner indicated in Fig. 14.
    One group of oscillations produces a single low-frequency variation (decrease) in the telephone current and a series of wave trains produces a corresponding series of low-frequency variations in the telephone current. In Fig. 15 is shown an oscillogram of the behavior of the audion when the "valve" action is employed for the detection of oscillations.
    With the means at hand it was impossible to ascertain the variations of the grid potential, as the leak introduced by connecting the oscillograph to the grid would destroy the cumulative action in the grid condenser. The grid potential, however, varies in exactly the same manner as the wing current. It will be seen that the fundamental detecting action is that of a valve, the high-frequency oscillations being rectified between the filament and grid, thereby causing a charge to accumulate on the grid and in the grid condenser. The charged grid then exerts a relay or trigger action on the wing current so that the audion is at once a rectifier and an amplifier. A somewhat similar combination of rectifying and amplifying actions occurs in the arrangement shown in Fig. 7. The action of the audion is being further studied by Prof. Morecroft and the writer in the research laboratory in electro-mechanics, Columbia University, and the results of these investigations will soon be published.