The valve is the active component in a vintage radio, doing all the really interesting jobs. Unfortunately, young engineers are no longer taught how valves work, so they have to teach themselves. These notes may help if you need to find out how a valve works.
In all atoms, there is a nucleus at the centre, with a cloud of electrons spinning around it. In a conductor, the electrons near the outside of the cloud are attracted to the nucleus by only weak forces. This means that if an electrical voltage is applied, the energy can slide electrons along from one atom to the next, and since electrons are negatively charged, an electron current flows from the negative terminal to the positive one. If the conductor is a good one, not much energy is required and the resistance to the flow of electrons is low.
If the conductor is heated, the electrons in the cloud attain greater velocities, and if the temperature is raised far enough, some electrons have enough energy to jump temporarily above the surface of the conductor. At this stage they are available to do something interesting. If not called upon, they will sink back into the material after a while. The amount of energy which the electron must have to escape from the material is called the "work function" of the material, and some materials (such as the Barium and Strontium oxides used for coating valve filaments and cathodes) have a reasonably low work function.
The phenomenon is known as thermionic emission, and hence the old fashioned texts called valves "thermionic valves". In a valve the emission is made to happen either by emitting them from an electrically heated wire filament (for battery valves which are directly heated) or by emitting them from a special coating on a metal tube which is warmed by an electrical heater within the tube. In this case the tube is called the cathode, and because the heater has no electrical connection to the rest of the circuit, it can be heated with AC, which saves the trouble of providing a nice smooth DC supply for the heaters when the set is working from AC mains.
Everything which follows works on the basis that the valve electrodes are contained in a vacuum, sealed up in their glass bottle, away from the air. Obviously if we are going to boil electrons out of a piece of metal and use them to do things, it will be no good if they just collide with gas molecules in the air and are wasted.
When the filament loses electrons, it becomes positively charged, because electrons are negatively charged. This means that it attracts the electrons back again. So of the electrons which boil off at all, only those with very high velocities will break right away. However, if we put a plate nearby with a positive charge on it, this will tend to attract the electrons. If the positive charge is sufficient it will draw quite a number of electrons towards it. Since this is a motion of electrically charged particles, a current is flowing.
This sort of valve has two electrodes, namely anode and filament, or if indirectly heated, anode and cathode. The symbol of the diode with a filament is shown below.
The diagram also shows the polarity of battery connections to observe a current flow. However, if an AC voltage source is used in place of the HT battery, the main use of the diode will become obvious. The diode will only conduct when the anode is more positive than the filament. When the AC input takes the anode negaitve, it repells all of the electrons back towards the filament and no current flows.
Hence we have passed all of the positive currents in the AC waveform and blocked all of the negative ones.
Looking at it another way, we have changed AC into DC, which is a useful thing to be able to do when we need to run a valve radio (which needs DC for the valve circuits to work) from the house mains supply, which give AC. We have obtained "lumpy" DC, as shown in the second graph below. This is better than we started with, but not much good really, as the radio circuits would only be working for half the time. The aim is to achieve proper DC, as shown in the third graph below. This can be approached by using the rectifier to supply the positive spikes of current to a large capacitor, which charges up while the supply is positive, and uses the stored charge to keep energy flowing when the rectifier is cut off on the nagative half-cycles. This is a reservoir capacitor. Every mains set has one!
We now commence to monkey around with the diode we have just made, and introduce a spiral of wire between the filament and the anode. We call this a grid. This opens up whole new possibilities. A basic demonstration circuit is shown below.
When the grid is positive, the electrons boiling off the filament are attracted away from it, and towards the grid. They will be accelerated through the space between filament and grid, and although some will impinge on the grid, constituting a small flow of current, some will pass through the gaps and be accelerated further until they impinge on the anode. Therefore a current flows from grid to filamant, and also a current flows from anode to filament.
If the grid is at zero volts or just slightly negative, it tends to push the electrons emitted back towards the filament. However, the anode is still positive and still attracts electrons. So some will be pulled away to the anode, and an anode current will still flow. They will be repelled from the grid on their way past, however, and so will not impinge on this and no grid current will flow.
As the grid voltage is made more negative, it becomes better at repelling the electrons back to the filament, and there is less chance for any to break away and head for the anode, although it is still positive. In the end a situation is reached when the grid is so negative that the anode current is cut off completely.
In practice, we do not use valves by taking the grid positive. This results in electrons impinging on the grid, which heats it up. Since it is a spiral of fine wire, in a vacuum, it has few ways of getting rid of the heat thus generated, and soon warms up. When it gets red hot, it starts to emit electrons of its own, which race away to the anode, and give a current from anode to grid which we were not expecting. Eventually the grid can melt. So in the first place we have got a current flow which we were not bargaining for, and in the end we have wrecked the valve.
Much more sensible, then, to use the valve with the grid voltage varied between cut-off and zero volts. Then the grid does not take current, and controls the anode current. Since the grid does not take any current, the input signal only has to have enough energy to alter the voltage on the grid in order to control the flow of current in the valve. Since the energy needed to control the grid is minuscule, we have a method of altering a reasonably sizeable anode current using a feeble input signal. This is amplification, and this is the crux of the matter.
Different types of valves have different internal construction, and this gives different characterisitics. One of the simplest characteristics is the relationship between anode current and grid voltage, which can be tested with the previous circuit. When the results are plotted on a graph, a typical result is shown below.
The three lines represent different anode voltages. Where the characterisitic curves are basically straight, the valve is giving linear amplification. In this region, a given change in grid voltage produces a fixed change of anode current. Outside this region, especially at very negative grid voltages near cut-off, the characteristics are curved, so a given change in grid voltage no longer produces the previously expected change of anode current. In this region, then, the valve is not responding in a well-predicted manner, and is distorting while amplifying. Hence, it's a bad idea to run the valve with the grid too negative, just as much as it is to run it with the grid too positive.
There are many ways of giving the valves a correct grid bais. There is the old-fasioned way of using a dry battery to bias the grid negatively with respect to HT-. However, grid-bias batteries were old fashioned even by the 1930s. The usual way is to make the filament a little positive with respect to HT-, which in most circuits can be arranged with just a resistor and some inginuity. The crafty way, however, is to use a grid-leak.
To give the grid good control of the valve, it is placed quite close to the filament. So when the valve is operating normally, some of the electrons emitted are drawn away to the anode, but most hang around between the filament and the grid in a little cloud of negative charge. This is called the space charge, and as some electrons fall back out of the cloud, into the surface of the filament, new ones are emitted. If the external circuit takes strong charge of the grid, the situation is well understood and the valve works as we have seen previously. However, if the grid is tied to HT- only by a high resistance, say 10 megohms, the space charge can begin to influence the grid voltage. With only a high resistance escape route, electrons from the space cloud impinging on the grid will tend to stay a while, and impart a negative charge to it. This then makes the grid go negative, and all we've had to do was to select the value of the resistor. It only works if the current in the valve is small, but you'll see this even in 1960s valve sets.
The triode is good, and it is useful, but it does throw up problems. The grid is easy to drive at low (audio) frequencies and everything is fine. But in any structure of metal things which are insulated from each other, it is possible to view the metal things as the plates and the insulator as the dielectric of a capacitor. So there are small capacitances inherent in the triode. These are from grid to filament and from grid to anode. Small capacitances have high impedances at low frequencies so the input signal to the grid is not unduly affected by them. However, at high frequencies, the impedances are smaller, and the feeble input signal will have to do work charging and discharging these parasitic capacitances. So at high frequencies a more powerful input signal is needed. But in a radio, the high frequencies are often straight from the aerial, and these are not powerful signals.
Of the two capacitances, one does more harm than the other. The capacitance between grid and filament is charged to the voltage of the input signal, by that input signal. But the capacitance between grid and anode is much harder work. In order to be made useful, the valve must have a load in its anode circuit to turn the amplified current into an amplified voltage. This means that as the anode current rises, the anode voltage falls. However, this means that as the input signal falls a little, the anode current drops a lot, and the anode voltage rises a lot. So the capacitance from grid to anode has to be charged not just to the signal input voltage, but to the signal voltage plus the output voltage. To charge a given capacitor to a higher voltage takes more current or more time, whichever is permitted. So we either need a more powerful input signal capable of driving current, or more time, which means we have limited the high frequency response of the valve.
This grid to anode capacitance is called the Miller capacitance, and has been understood since the 1920s. The key to making better high-frequency valves was to find a way around it.
The solution which worked first was the screen-grid valve. In this valve another grid was added, between grid and anode. This was the screen grid. It was beefier than the control grid, and was connected to an HT voltage supply rail of a voltage lower than the anode voltage. In terms of electrons, it accelerated the electrons away from the grid and out past itself where they were then attracted to accelerate towards the anode. So it did not make a great improvement to low frequency operation.
However, the predominant capacitances from the control grid were now from grid to filament and grid to screen grid. Since neither the filament nor the screen grid changed voltage in sympathy with the signal, the effect of the Miller capacitance was greatly reduced and higher frequencies could be handled better.
There was a problem, however. The anode volts vs. anode current characteristic curve had a pronounced kink in it. When the electrons hit the anode, they knocked some electrons out of the surface of the metal making up the surface of the anode, and these could be attracted back to the screen grid, if they came out with enough velocity. This gave less anode current than expected, because the anode had spilt some electrons while accepting others.
When the anode voltage was very low, the attraction for electrons was small, so current was small. As the voltage increased the attraction for electrons grew and so did the current. Then we reach the point where the electrons impinging on the anode have enough energy to dislodge other electrons which spill out back to the screen grid. So the anode current falls. Then as the anode voltage rises further, the current rises away again.
Having a kink in the anode characteristic like this meant that the valve would distort a large input signal very badly if it made the valve run over this kinked region of its characteristics. So it was no use for large audio signals. It was useful for small radio frequency signals, however, as long as it was biassed so that it was working well away from the kink.
It paid to be careful with the screen-grid valve even when using it for its intended purpose. The kink in the anode characteristic meant that over a small range of voltages, an increase of voltage produced a decrease in current. You can think of that as negative resistance, which is a rare thing indeed. However, a valve baissed up like this with a tuned circuit in its anode circuit would oscillate for sure. It's called a Dynatron oscillator. And if the bias was messed up, you might have ended up with one!
The Screen-Grid valve was an advance but there was a quite justifiable feeling that it was a half-finished project, and mankind could do better. This was indeed the case. The addition of another grid, between the screen grid and the anode, made a pentode.
The third grid was called the suppressor grid, and it is usually tied to the filament. This means that electrons which spill out of the anode cannot get back to the screen grid. They are repelled by the suppressor and fall harmlessly back to the anode, where they put back the charge which so nearly ran away, and the nasty kink of the tetrode is eliminated.
Pentodes can be used for large audio signals, and they have been used as audio output valves since the late 1920s. Compared with triode output valves they are more sensitive, needing less drive signal voltage to the grid for full whack output, which makes them easier to use. They don't sound as nice as triodes in their raw state, but can be put into specially engineered output stages to approach the sound quality of triodes.
The other advantage of a pentode is that it can be engineered to give more gain per stage than triodes in small signal applications. This is useful in many circumstances, either where the input signals are small, or where feedback is being used, and extra gain is needed to obtain the required overall high gain and low distrotion from the amplifier as a whole. For low level input signals, care is required, however, because the pentode is inherently noisier than a triode.
You can go mad with this grid business. Octodes and Nonodes are known. The ultimate aim of all of this oddball gear is usually to make a valve which can be controlled by two input signals.
The reason for this is that a Superhet radio, which most radios are, relies on the availability of a valve which can mix two signals together. In its crudest form, this can be just a case of making one signal distort the other very much, and this is the usual course of action with frequency changer valves.
The aim is to mix the tuned-in incoming signal from the transmitter with a signal from an oscillator inside the set, tuned to a frequency offset from the tuned-in signal by a fixed amount (often 465kHz). The resulting output is a right old mess, containing components of signal frequency, oscillator frequency, but more importantly, the offset frequency (the magic 465kHz) to which the rest of the gear in the radio is permanently fixed-tuned. This means that the rest of the gear in the set can be as complicated as you like, but you only have to tune it all in to 465kHz and then leave well alone, as long as the oscillator stays 465kHz from the received signal, and the frequency changer keeps working.
The most common sort of valve for frequency-changer use is a triode-hexode. The triode is the oscillator and the hexode is the mixer. For the hexode, the tuned-in station goes to the control grid, and the oscillator signal is connected to (usually) the third grid, where it chops up the signal from the control grid as it is amplfied in the valve, performing the mixing operation. Usually grids 2 and 4 are connected to HT, like the screen grid in a pentode.
More complicated valves are normally an attempt to get the oscillator and the mixer into the same electrode structure, and the many-grid valves like 6BE6, 6A8, etc, are almost all designed to be self-oscillating.
So that's what's in the bottle. What goes wrong with it?
Loss of Emission.
The process of wearing out. Eventually the filament or cathode fails to emit electrons at the required rate. Precious little you can do about this one. New bottle required.
Gas-Leak. (Going Soft)
The bottle cracks and air gets in. The silvery "getter" on the glass goes white and peels off. Terminal. New bottle required.
Or, of course, the more subtle and nasty version, where just a tiny amount of gas gets in. The valve glows purple inside when working, runs huge currents, gets extremely hot, and if you're lucky, damages the radio. New bottle required.
In a mains valve with a heater and a cathode, the insulation between heater and cathode can break down, so the AC heating the valve gets into the circuit proper and makes it all hum.
Not much you can do, normally. If the valve is impossible to replace, imagination required. DC heater feed? Easy option is normally replacement.
Not common, but occasionally bits of the internal metal structure drop off and rattle around causing random short circuits inside the valve. Can give nice sparks and damage radios. Discard valve and find a whole one.
Not that common. Bits come disconected. Often it's the cathode which comes disconnected because there has been a fault causing a huge overload of cathode current to flow. The connction inside the valve will act like a fuse. Other bits can come unwelded and thereby disconnect themselves. Replace.
Tap the valve. Loud rings and other noises in the speaker. Nightmare. Perfectly good valve which causes trouble whenever it's vibrated. You can try mounting the valveholder in rubber floats, or wrap rubber bands around the valve. Or you can save your time and sanity and just order a new one.
Crackles, pops, hissing, they can do it all. Replace it. It won't get better.