User:I B Wright/sandbox

Not necessarily the final version but good enough for discussion

How the audion works
The audion is a triode device in that it featured three electrodes, but it is quite distinct from the vacuum triode. It should not be necessary to refer to an audion as a triode because there were never any audions with more electrodes, rendering it unnecessary to qualify the name with the number of electrodes. However, De Forest himself quickly did so because of his belief (not to mention vested interest) in claiming that the vacuum triode was the same device as his audion. He was wrong on the point, but then De Forest himself did not know how his audion worked let alone the vacuum triode.

Readers should at least be familiar with how the vacuum triode works. Briefly, the incandescent filament emits electrons which are attracted to the positively charged anode which passes to the external circuit as anode current. The grid, located between the filament and the anode is biased with negative voltage with respect to the filament and repels some of the electrons back to the filament. If this negative voltage is reduced, fewer electrons are repelled and the anode current increases. Similarly, if the negative voltage is increased, more electrons are repelled and the anode current decreases.

The internal operation of the audion is much more complex, and will be broken down into sections. The term ‘vacuum triode’ is used here to distinguish that device from the audion. The term audion itself was almost a portmanteau of ‘audio-‘ and ‘-ion’, betraying that the device relied on ion current flow for its operation.

In order to avoid confusion, any reference to current flow is not necessarily used in its conventional sense (i.e. positive to negative). References to electron current refers to the direction of electron flow (i.e. negative to positive) and references to ion current flow refers to the direction of the positively charged ionised gas molecules (i.e. positive to negative).

Anode circuit
In the same manner as in the vacuum triode, the filament emits electrons by thermionic emission. Also similarly, these are attracted to the positively charged anode. This is where the similarity ends. In the audion, some of these electrons collide with the gas molecules present in the device. This collision ‘knocks’ additional electrons from those molecules leaving positively charged ions which form an ion current flow from the anode to the filament. The liberated electrons contribute to the electron current flow and in turn some collide with more gas molecules contributing further to the ion current flow. This is not a runaway condition and the overall anode current will settle to a steady value. Internally, the anode current consists of negatively charged electrons (electron current) flowing from filament to anode and positively charged ions (ion current) flowing from anode to filament, both of which contribute to the total anode current flowing in the external circuit. The presence of the gas molecules effectively magnifies the anode current that would otherwise flow.

It is not possible in any device to determine the magnitudes of the electron current or the ion current or indeed their magnitude relative to each other. These magnitudes depend on the exact pressure of the gas in the tube. Because of the interaction of electrons and gas molecules, coupled with the absence of a negative grid bias, the audion operates at a much lower anode voltage than a similar vacuum triode. Anode voltages of between 20 and 25 volts were normal for De Forest’s audion tubes. A similar vacuum triode of the time typically required 120 volts on its anode.

Grid circuit
Consider at this stage that the grid has no external circuit connected. By virtue of its juxtaposition between the filament and the anode, some of the electron stream collides with the grid wire. Similarly, part of the ion stream flowing the other way also collides with the grid. Because there is no external connection, it follows that the electron current to the grid matches the ion current yielding a net cancellation. The exact magnitude of the two currents is known as the ‘grid crossover current’. By virtue of these currents, the grid will assume a potential relative to the filament. This voltage is known as the ‘grid crossover voltage’. In an ideal audion it would be zero volts relative to the filament. An increase in gas quantity will result in the grid crossover voltage becoming a little more positive and similarly a reduction will result in it becoming a little more negative. In the range of gas pressures encountered in production audions, any consequent change in grid crossover voltage is virtually negligible.

In any audion device, it is not possible to determine the magnitude of the grid crossover current because it is entirely internal to the device. It is also not easily possible to directly determine the grid crossover voltage, but this latter characteristic can be determined indirectly. One simple method is to apply a small variable voltage to the grid and adjust it such that the anode current is the same as when it the grid was open circuit.

Dynamic operation
When an alternating signal is applied to the grid it causes a current to alternately flow into and out of the grid. When electrons flow out of the grid connection there is now a net imbalance of electron current and ion current striking the grid. The increased grid electron flow reduces the electrons contributing to the anode current and hence available to ionise the gas molecules and this in turn reduces the ion current flow. In this manner, once again the presence of the gas has a magnifying effect on the influence of the grid current. In the same manner, when electrons flow into the grid, there is once again a net imbalance of ion and electron currents striking the grid, but this time more electrons are now available to the anode current and to ionise the gas and the ion current increases. It should be noted that at no time can there be a net electron flow passing from the grid into the device, because the grid cannot emit electrons as it is not heated. Any electron flow into the grid is cancelled by the ion flow within the device. Just as in the case of the quiescent condition, the presence of the gas also has a magnifying effect on the influence of the grid current.

This interaction provides a very important difference between an audion and a vacuum triode. The audion is a current operated device (that is, current flow in the grid circuit is essential to affect the anode current flow). Contrast the vacuum triode which is a voltage operated device, because at the negative grid bias required for normal operation, no current flows in the grid circuit. This difference also means that although the vacuum triode is a very high impedance device (input impedances are of the order of tens to hundreds of Megohms), the audion is a medium impedance device (input impedances are of the order of a few kilohms). A radio detector with even that medium impedance was unknown prior to the invention of the audion.

Linearity
The ratio of the change in anode current change to grid current (the gain) is dependent upon the relative magnitude of the anode ion current to the anode electron current. The effect on the electron current flow is more significant than the effect on the ion flow, because the mass of the ions are approximately 27,000 times greater than that of the electrons. This means that as the electron current reduces, the ion current also reduces but by a smaller amount making the ratio of electron to ion current smaller. Similarly, as the anode electron current increases, the ion current also increases but again by a smaller proportion, making the ratio of electron to ion current greater. This relationship means that the amplification that the device provides to the grid current is distinctly non linear.

A non linear characteristic is exactly what is required to detect radio signals, and as a bonus the device amplifies the remaining demodulated signal. But with a non linear characteristic, the device is all but useless to amplify any signal further.

Push-pull operation
It had been figured out that it might have been possible to operate two audion tubes in a push pull arrangement, with the non linear characteristics cancelling each other out (and some early patents showed this arrangement, though made no specific claims). Unfortunately, to make this work means that the two audions have to be matched such that they have the same grid crossover current. Since the currents that contribute to the crossover currents are internal to the device, it was not possible to determine or compare them.

Frequency response
The anode and grid currents are composed of both electron and ion current flow. Since the ions have approximately 27,000 times greater mass than the electrons they move more slowly and are much harder to accelerate. This greatly limits the frequency response of an audion which exhibits a bandwidth of around a fifth that of a similar vacuum triode.

Failure mode
The primary failure mode of the audion is not failure of the filament, which was the case with early vacuum triodes. Because the device relies on a filament heated to incandescence, the filament in turn heats the glass bulb, the grid and the anode. This heating causes these parts to adsorb the gas molecules within the device. As the molecules are adsorbed, they are removed from the electron current’s path. Consequently, the grid crossover current reduces as there are fewer ions to contribute to it. Further, the magnifying effect afforded by the gas is reduced as gas molecules are removed from the active parts of the device. Also because there is now a net surplus of electrons colliding with the grid, the grid crossover voltage becomes substantially more negative. This negative voltage starts to repel electrons back to the filament (as the device proceedes to operate more like a vacuum triode and less like an audion as the failure progresses). This progressively reduces the anode current. Eventually, the grid becomes negative enough that it very nearly cuts the anode current off, effectively stopping the device from working completely.

Electrometer
A more modern device that exploits the principles of the audion is the electrometer. Electrometer tubes are constructed substantially as vacuum triodes with as much gas removed as possible before sealing and then subsequently gettered to remove as much residual gas as possible. However, electrometers are designed to operate with very tiny currents indeed. Most electrometers operate with an anode voltage of just 1.4 volts. At this anode voltage, the anode current is just a few hundred picoamps. At this very low current, what residual gas there is in the device is sufficient that the anode current consists of both electron flow and a substantially similar ion flow, albeit both extremely small. As a consequence, the device is current operated in the same way that the audion is, but the grid current required to produce an appreciable change in anode current is extremely minute. Some electrometers were constructed to amplify currents as low as a few femtoamps (a femtoamp being 10-15 amps). Like the audion, the electrometer is non linear in operation and, once again, numerous attempts have been made to cancel the non linearity by operating them in a push-pull arrangement. However, the continued inability to match them for grid crossover current thwarted any such attempts. Unlike the audion, the electrometers do not fail to gas adsorption, because the very small amount of gas is not easily adsorbed and, in any case, is replaced due to leakage which is controlled by the getter.

Electrometers survived as amplifiers of such tiny currents to well towards the end of the 20th century, but advances in MOSFET technology have now supplanted their use.

Conventional valves
It is quite feasible to operate a conventional vacuum tube at current levels where the residual gas provides the method of operation necessary to amplify very small grid currents. The very small amount of gas would not normally significantly affect the operation of such a valve when used in its intended manner. The Brimar company published a circuit for a valve based 0 to 5 picoamp meter in 1968. It did not use a valve type specifically constructed as an electrometer. It did use a standard pentode small signal valve, the 6BS7. In this circuit, the valve had no grid bias and thus worked by using the grid current to modulate the anode current of just 0.6 microamps (the device normally operating with an anode current of 2 milliamps). The non linear characteristic of silicon diodes were used in an attempt to compensate for the non linearity that would have resulted from the audion like operation of the 6BS7, but the circuit note stated that one resistor had to be adjusted for best linearity (implying that the circuit would never be completely linear).

Summary of differences
The audion requires low pressure gas for its operation. The vacuum triode requires a hard vacuum. A vacuum stops an audion from working. Any present gas will stop a vacuum triode from working.

The audion operates at much lower anode voltages than an equivalent vacuum triode. Required voltages are 15 to 20% that of a vacuum triode.

The audion is a current operated device (i.e. it requires grid current to operate). The vacuum triode is a voltage operated device (i.e. there is no grid current).

The audion is a medium impedance device with an input impedance of typically a few kilohms. A vacuum tube is a very high impedance device with an input impedance of a few hundred Megohms.

The audion fails quickly due to gas adsorption, ironically ultimately becoming a vacuum tube.

The audion operates entirely without any bias on its grid circuit. The vacuum triode, conversely, requires a steady negative bias on its grid for it to operate correctly.

The audion is a distinctly non linear device (which was useful for radio signal detection). The vacuum triode is substantially a linear device which initially made it useless for radio detection, but very useful for subsequent amplification of a detected signal. Techniques were later developed enabling vacuum triodes to detect radio signals (essentially by operating them on a non linear portion of their characteristic).