Multigrid Vacuum Tube Theory

Preface

This document presents only a very brief description of how vacuum tubes operate in general, concentrating on the purposes of the screen grid and suppressor grid, explaining why they were introduced and what functions they serve. Also presented is a brief explanation of how beam power tubes operate. For a more detailed presentation of the general operation of vacuum tubes, see the author's document "Tube Theory", available on the VRPS web site. Not covered is the functioning of special purpose multigrid tubes, like the pentagrid converter and triode/hexode converter tubes. These multigrid tubes deserve a presentation of their own. Also not covered is "oddball" uses of the additional grids outside of their intended use, as in screen modulation, etc.

Introduction

As described in "Tube Theory", the cathode of a vacuum tube emits electrons into the vacuum chamber, creating a space charge. The plate of the tube attracts these electrons, resulting in a current flowing through the vacuum chamber. The control grid intercepts this flow, and a negative voltage on this grid reduces the current available at the plate. Thus the voltage on this grid can be used to control the plate current, hence the name "control grid".

This would seem to be the all one needs to know about tubes, since control of the current in the plate is all that one needs, and indeed that seemed to be the case when audio frequency signals were being amplified. However, when tubes were used at higher frequencies, and at high powers, some unanticipated limitations were encountered. The screen grid was introduced to improve the performance at high frequencies, and the suppressor grid and beam power tubes were developed to allow higher plate voltages and powers to be used. As it turned out, the screen grid also allowed tubes to have much higher gain.

Triodes

How They Function

Triodes are "tri", hence three, element tubes. They are the simplest tubes which are capable of amplification (and hence oscillation). The three elements are the cathode, which emits electrons, the plate, which collects electrons, and the grid, which limits how many electrons reach the plate. The voltage on the grid controls the current flowing out of the plate, the more negative the voltage on the grid, the less the plate current. Minute grid voltage variations can result in dramatic variations in plate current, and negligible grid power can control immense plate power. This is the basic way all tubes, regardless of the number of grids, operate. Why were the other grids added?

An Unexpected Limitation

At first, when tubes began to be used as amplifiers, they were simply attached to a crystal receiver, and used to amplify the detected audio signal for purposes of driving a loudspeaker. While this greatly increased the volume available for listening, and made group listening a much more enjoyable experience, it did very little toward increasing the sensitivity of the receiver. If the radio frequency (RF) signal were below the thresh hold of the detector, there would be no detected audio to amplify. If the sensitivity could be increased, it would be possible to listen to more distant stations. However, when designers attempted to use triodes to amplify the RF signals, they encountered difficulties in the form of unwanted oscillations.

As anyone who has been present when a public address (PA) amplifier was being used knows, when the output of an amplifier gets back into the input (feedback), the amplifier becomes an oscillator. In a PA amplifier, this results in howls and squeals. In RF equipment, the oscillations are not audible, but they are just as real, and just as objectionable. At audio frequencies, triodes performed quite well. However, when triodes were first used as RF amplifiers, they promptly became oscillators. It seemed that amplification of RF signals was impossible. Even after all external feedback paths were eliminated, the tubes still insisted on oscillating. It turned out that the feedback path was inside the tube itself. The neutrodyne circuit, developed to overcome this obstacle, included external circuitry which introduced feedback exactly opposite to that occurring inside the tube to compensate for this limitation. Unfortunately, the compensation circuit needed fiddly adjustment for the individual tubes used, and provided exact cancellation only at one frequency. This was not an acceptable long term solution.

Why Do They Oscillate?

The problem is caused by internal feedback inside the tube. The feedback is a result of the capacitance between the output electrode (plate) and the input electrode (control grid). During normal operation of the tube, the plate voltage, of course, varies. When the control grid voltage is made more positive, the plate current increases, resulting in a lowering of the plate voltage, that is, the plate becomes more negative. In this way, the voltage on the plate, though it reproduces the waveform of the signal, also reverses its phase. At high frequencies, the inter-electrode capacitance between the plate and the grid couples some of the output voltage back to the grid.

If some of the output signal from an amplifier gets coupled (fed back) into the input, then part of the output becomes a part of the input signal, and the amplifier treats it as such. If the output signal is out of phase with the input signal, then they cancel somewhat, and although the effective gain is reduced, nothing untoward occurs. In fact, there are some potential benefits of this, and this so-called "negative" feedback is sometimes used deliberately, and one trades off gain for other benefits. However, if the signal fed back is in phase with the input, then the input signal is augmented, and a portion of the output signal gets amplified again, resulting in increased effective gain. If enough of the output gets fed back in phase, then the output signal can actually be all the input signal the amplifier needs to produce a signal on its own. Any small input signal, like noise (which is always present) can cause this. This is what happens when a PA microphone can "hear" the sounds coming from the loudspeakers. An oscillator may be defined as being a device which provides an output signal without the need for an input signal.

It seems that any feedback from the plate of a tube to its grid would be exactly out of phase with the input signal, and would not cause a problem. However, there is an effect called "transit time". When the voltage on the control grid changes, it takes a little time for the electrons passing the grid to reach the plate, so there is a delay before the effect on the electron stream affects the voltage on the plate. Also, if there is any capacitive load on the plate it takes a while for the current change to alter the voltage on that capacitor, which must be charged or discharged somewhat. Other reactive loads, like tuned circuits, can result in variable delays depending upon their tuning and also upon frequency. In any case, there will be a frequency at which the delay is precisely one half cycle of a signal at that frequency. If at that frequency the amplifier has enough gain, and if there is enough feedback present in the circuit, the amplifier will oscillate, without the need for any input. In the PA example, most of the delay comes from the time it takes the sound to get from the speaker to the microphone.

Screen Grid Tubes

A Better Solution...

There had to be a better way than adding external feedback components requiring fiddly adjustments. What was required was a way to suppress the internal feedback inside the tube, instead of trying to counteract it by using external components. About 1926 Dr. A. W. Hull added a screen grid, creating the tetrode, "tetra" being the greek word for four, and "-ode" from "electrode". The purpose of the screen grid was to interpose a shield between the plate and the control grid of the tube to suppress the internal feedback. The screen, though between the plate and the control grid, could not be a solid structure, because it had to permit the electrons to pass through. Dr. Hull came up with the idea of another grid, made from a very open structure, i.e., wound much less tightly than the control grid, so it would not intercept much of the electron stream passing by it.

This grid would need to be at a positive voltage, or it would repel the electrons and impede the electrons from reaching the plate. However, it needed to be at signal ground to screen the control grid from the capacitance of the plate. As a consequence, the screen was usually fed from the B plus supply by a resistor, and bypassed by a capacitor to the cathode. The resistor and capacitor acted as a filter keeping the RF from the power supply, and also keeping the screen at signal ground (though at a high DC potential). Another feature of the screen grid was that it was connected to a (usually perforated) cylindrical shield surrounding the plate, thus acting as an internal screen for the tube as a whole, somewhat protecting it from unwanted external feedback, and from radiating its own interference to other tubes. This dual structure of the screen grid can be seen in early schematic diagrams where the screen grid is shown going around both sides of the plate. The new screen grid electrode provided an entirely satisfactory solution to the problem of internal feedback within triodes, and eliminated the need for additional components and tricky adjustments.

Another benefit of the screen grid tube is that the screen grid is actually the anode of the tube, and the cathode, control grid, and screen grid form a sort of triode with the triode plate (screen grid) grounded for signal purposes (by bypassing), shooting electrons at the plate, which becomes merely a target. In this way the current in the plate is nearly independent of the plate voltage. This allows one to obtain a much greater gain (g_m) from a single tube than is possible with triodes.

The amplifier as a whole still needed shielding, especially around the transformers and even individual tubes, but this was easy to supply in the form of sheets of metal, which do not require adjustments. Also, the antenna and its lead in wire needed to be separated somewhat from the receiver to prevent this feedback route, but this was easy to accomplish. Since the internal capacitance was reduced to negligible amounts, the maximum frequency at which tubes could be used was markedly increased. There were two reasons for this. First, of course, the internal feedback at high frequencies was eliminated. Additionally, the capacitive load on the output was reduced. This load increases with frequency. So, there were five benefits to the screen grid tubes: elimination of internal feedback, reduced capacitive load on the plate, reduced susceptibility to external interference, reduced production of interference with other tubes, and increased gain. All in all, quite a success!

...But Not a Perfect One

Along with the benefits of the screen grid, came some less beneficial properties of the tubes. The first noticeable undesirable property was increased noise, when compared to triodes. The reason for this is that we now have two possible destinations for electrons emitted by the cathode; they may exit via the plate, or they may exit via the screen. Though the screen grid is a very open structure, and only about ten per cent of the electrons reach the screen grid, still, some do reach it, and the number of electrons apportioned to the plate versus the screen is random over time. This constitutes a noise current (partition noise) in the tube, which is unavoidable in tubes with more than one positive electrode. Other multigrid tubes, like the pentagrid converter, for example, have this problem. The more positive grids there are in a tube, the worse the problem is.

Another undesirable property of screen grid tubes noticed early on is that, while the voltage on the screen grid is maintained constant via bypassing, the voltage on the plate is not. When used with strong signals or at high powers, the plate voltage may fall below that of the screen, and electrons which actually made it past the screen going toward the plate might reverse course and go back to the screen. In extreme cases, this could lead to electrons circling around the screen repeatedly before finally striking it, leading to oscillations in the screen grid circuit if the screen bypass wiring was excessively long. Eventually, this effect was used to create extremely high frequency oscillators (dynatrons), but until the cause was understood, the source of these high frequency oscillations was a bit of a mystery, and they were considered to be a nuisance in normal circuitry.

A third undesirable property of screen grid tubes, also a mystery at first, was the so-called "tetrode kink" in the plate characteristics of the tube. High power tubes were especially susceptible to this feature. It was found that at high currents, when the plate voltage was lower than that on the screen, the plate current could actually decrease when the plate voltage was increased. In other words, increasing the B plus supply to the plate could decrease the current in the plate, and making the control grid voltage more negative could result in an increase in the plate current, rather than decreasing it as expected. In some extreme cases, the plate current could actually reverse with the plate supplying more current to the screen than it received from the cathode. The plate current in the type 24A tube is definitely negative over part of its curve.

The mechanism for this somewhat perplexing situation is called "secondary emission", in which the plate, not intended to emit electrons, actually does. The way this happens is that electrons striking the plate could dislodge other electrons from it, scattering them into the space charge, similar to the "break" in a game of billiards. These scattered electrons find themselves in the vicinity of a plate which is more negative than the screen, and hence head for the screen. An incident electron could dislodge two, or even more electrons from the plate, resulting in a net current flowing out of the plate and into the screen. This is not an issue with triodes, since any secondary electrons would be attracted back to the plate, and exit thereby. Various schemes were tried to reduce this effect, including coating the plate with graphite, but nothing completely eliminated the problem.

Some way was needed to get these secondary electrons back to the plate.

Pentodes and Beam Power Tubes

Not long after Dr. A. W. Hull added the screen grid, possibly in the same year, he also added yet another grid, the suppressor grid. This grid, even more open than the screen grid, is intended to be maintained at a fixed negative potential, which would always be negative with respect to the plate. It was usually connected to the cathode, in some tubes actually inside the tube itself. Unlike the screen grid, this grid can have a negative charge, because the electrons have already been accelerated by the screen. The tetrode no longer relies upon the positive voltage on the plate to be sensible to electrons near the cathode. The electrons do indeed slow down somewhat as they pass the suppressor grid, but this is only a "road bump" in their journey, not a barrier. The grid's name comes from the fact that it suppresses the flow of secondary electrons from the plate to the screen grid, repelling them back to the plate, causing them to exit thereby. This allows tubes to be used at high currents where the plate voltage would be low, and also at high powers where electrons might strike the plate very forcibly, ejecting more electrons. The benefits were twofold: elimination of the tetrode kink, hence reducing distortion, and increased power handling capability. These tubes, with five electrodes, were called "pentodes", "penta" from the greek word for five, and "-ode" again from "electrode".

When the european tube manufacturers saw the advantages of the pentode for use in high power audio output tubes, they wanted to produce them, but due to patent licensing issues, found themselves unable to do so profitably. As a consequence, they sought to find another way to return the secondary electrons to the plate, and indeed came up with another successful design, called the "kinkless tetrode". To this day, european tubes using this principle have assigned numbers starting "KT", like the KT88. In the United States, they were called "beam power tubes".

There were actually three different changes made to the design of tubes to create the beam power tubes. The first departure was the addition of what were called "beam forming electrodes". These were electrodes introduced around the screen grid, and between it and the plate, which electrodes were in the shape of parentheses, and were connected to each other and to the cathode. These electrodes were near the support wires which fixed the screen grid in place. They had the effect of compressing the electron stream into two intense beams directed at the plate in two diametrically opposite areas. Another consequence was that the electrons slowed down as they entered the compressed region, both because of the effect of the negative charge on the beam formers, but also because of the increased space charge of the dense beam. This resulted in a sort of bow wave of negative charge between the beam formers and the plate. This increased density of space charge repelled the secondary electrons in a manner similar to that of a suppressor grid, since it provided a region of space which was more negative than the plate. This effect was nearly as effective as an actual suppressor grid.

The second modification to tube design was to align the screen grid wire turns with those of the control grid. This had the effect that the screen grid intercepted a lesser percentage of the total electron stream, since the stream, avoiding the turns of the negative control grid, also tended to avoid the screen grid. This had two beneficial effects. First, it reduced the power handling requirements of the screen grid, which was a not inconsequential thing in a tube intended to handle large powers, and second tended to reduce somewhat the partition noise current in the plate.

The third change was to change the shape of the cathode, the grids, and the plate from being circular cylinders to being elliptical cylinders, with the beam forming electrodes near the vertices of the ellipses. This provided a flatter region in space for the space charge, and also allowed for some surface area of the plate not intercepting the beams to help radiate away some of the heat generated by the intense beams. This reduced the kink in the plate curves even further.

Summary

We see that when triodes were first used to amplify RF signals, unanticipated oscillations resulted from the unavoidable internal feedback inside the tubes. When the screen grid was introduced, it eliminated this problem and allowed much higher frequency operation and much higher gain, but brought with it the problems of partition noise current, and distortion in the plate characteristics. The distortion was solved by the introduction of yet another grid, the suppressor grid, for use in low power RF tubes, and either a suppressor grid or beam formers for use in high power audio tubes.