Electronic amplifiers are built up from combinations of active and passive components. The active ones are those, like tubes or transistors or integrated circuits, that draw electrical current from suitable voltage supply lines and then use it to generate or modify some electrical signal. The passive components are those, like capacitors, resistors, inductors, potentiometers or switches, which introduce no additional energy into the circuit, but which act upon the input or output voltages and currents of the active devices in order to control the way they operate. Of these, the active components are much more fun, so I will start with these.
Although the bulk of modem electronic circuitry is based on 'solid state' components, for very good engineering reasons -- one could not, for example, build a compact disc player using tubes, and still have room in one's house to sit down and listen to it- all the early audio amplifiers were based on tubes, and it’s useful to know how these worked, and what the design problems and circuit options were, in order to get a better understanding of the technology. Also, there is still an interest on the part of some 'Hi- Fi' enthusiasts in the construction and use of tube operated audio amplifiers, and additional information on tube based circuitry may be welcomed by them.
"Valves" or "Vacuum Tubes"?
The term thermionic valve (or valve for short) was given, by its inventor, Sir Ambrose Fleming, to the earliest of these devices, a rectifying diode. Fleming chose the name because of the similarity of its action, in allowing only a one-way flow of current, to that of a one-way air valve on an inflatable tire, and the way it operated was by controlling the internal flow of thermally generated electrons, which he called 'thermions', hence the term thermionic valve. In the USA they are called 'vacuum tubes' (or just ‘tubes’). These devices consist of a heated cathode, mounted, in vacuum, inside a sealed glass or metal tube. Other electrodes, such as anodes or grids are then arranged around the cathode, so that various different functions can be performed.
The descriptive names given to the various types of tube are based on the number of its internal electrodes, so that a tube with two electrodes (a cathode and an anode) will be called a 'diode', one with three electrodes (a cathode, a grid and an anode) will be called a 'triode', one with four (a cathode, two grids and an anode) will be called a 'tetrode', and so on.
It helps to understand the way in which tubes work, and how to get the best performance from them, if one understands the functions of these internal electrodes, and the way in which different groupings of them affect the characteristics of the tube, so, to this end, I have listed them, and examined their functions separately.
The cathode
This component is at the heart of any tube, and is the source of the electrons with which it operates. It’s made in one or other of two forms: either a short length of resistor wire, made of nickel, folded into a 'V' shape, and supported between a pair of stiff wires at its base and a light tension spring at its top, as shown in FIG. 1 a, or a metallic tube, usually made of nickel, with a bundle of nickel or tungsten heater wires gathered inside it, as shown in FIG. 1b. Whether the cathode is a directly heated 'filament' or an indirectly heated metal cylinder, its function and method of operation is the same, though, other things being equal, the directly heated filament is much more efficient, in terms of the available electron emission from the cathode in relation to the amount of power required to heat it to its required operating temperature (about 775 degr. C for one having an oxide coated construction).
FIG. 1 Tube cathode styles (a) Directly heated (b) Indirectly heated [Cathode
connection ]
It’s possible to use a plain tungsten filament as a cathode, but it needs to be heated to some 250 degr. C to be usable, and this requires quite a substantial amount of power, and leads to other problems such as fragility. Virtually all contemporary low to medium power tubes use oxide coated cathodes, which are made from a mixture of the oxides of calcium, barium, and strontium deposited on a nickel substrate.
In the manufacture of the tube these chemicals are applied to the cathode as a paste composed of a binding agent, the metals in the form of their carbonates, and some small quantities of doping agents, typically of rare-earth origin. The metal carbonates are then reduced to their oxides by subsequent heating during the last stage in the process of evacuating the air from the tube envelope.
In use, a chemical reaction occurs between the oxide coating and the heated nickel cathode tube (or the directly heated filament) which causes the alkali metal oxides to be locally reduced to the free metal, which then slowly diffuses out to the cathode surface to form the electron emitting layer. The extent of electronic emission from the cathode depends critically upon its temperature, and the value chosen for this in practice is a compromise between performance and life expectancy, since higher cathode temperatures lead to shorter cathode life, due to the loss through evaporation of the active cathode metals, while a lower limit to the working temperature is set by the need to have an adequate level of electron emission.
When hot, the cathode will emit electrons, which form a cloud around it, a situation in which the thermal agitation of the electrons in the cathode body, which causes electrons to escape from its surface, is balanced by the growing positive charge which the cathode has acquired as the result of the loss of these electrons. This electron cloud is called the 'space charge', and it plays an important part in the operation of the tube; a matter which is discussed later.
The anode
In the simplest form of tube, the diode, the cathode is surrounded by a metal tube or box, called the anode or plate. This is usually made of nickel, and it will attract electrons from the space charge if it’s made positive with regard to the cathode. The amount of current which will flow depends on the closeness of the anode box to the cathode, the effective area of the cathode, the voltage on the anode, and the cathode temperature. For a fixed cathode temperature and anode voltage the ratio of anode voltage to current flow determines the anode current resistance, R_a, which is measured by the current flow for a given applied voltage -- as shown in the equation
R_a = dV_a/dI_a
Because the anode is bombarded by electrons accelerated towards it by the applied anode voltage, when they collide with the anode their kinetic energy is converted into heat, which raises the anode temperature. This heat evolution is normally unimportant, except in the case of power rectifiers or power output tubes, when care should be taken to ensure that the makers' current and voltage ratings are not exceeded. In particular, there is an inherent problem that if the anode becomes too hot, any gases which have been trapped in pores within its structure will be released, and this will impair the vacuum within the tube, which can lead to other problems.
The control grid
If the cathode is surrounded by a wire grid or mesh - in practice, this will usually take the form of a spiral coil, spot-welded between two stiff supporting wires, of the form shown in FIG. 2 -- the current flow from the cathode to the anode can be controlled by the voltage applied to the grid, such that if the grid is made positive, more negatively charged electrons will be attracted away from the cathode and encouraged to continue on their way to the anode. On the other hand, if the grid is made negative, it will repel the electrons emitted by the cathode, and reduce the current flow to the anode.
FIG. 2 Control grid construction. Note: Support rods
It’s this quality which is the most useful aspect of a tube, in that a quite large anode current flow can be controlled by a relatively small voltage applied to the grid, and so long as the grid is not allowed to swing positive with respect to the cathode, no current will flow in the grid circuit, and its effective input impedance at low frequencies will be almost infinite. This ability to regulate a large current at a high voltage by a much smaller control voltage allows the tube to amplify small electrical signals, and since the relationship between grid voltage and anode current is relatively linear, as shown in FIG. 3, this amplification will cause relatively little distortion in the amplified signal. The theoretical amplification factor of a tube, operating into an infinitely high impedance anode load, is denoted by the Greek symbol u (“μ”).
Although in more complex tubes there may be several grids between the cathode and the anode, the grid which is closest to the cathode will have the greatest influence on the anode current flow, and this is therefore usually called the control grid.
The effectiveness of the grid in regulating the anode current depends on the relative proximity of the grid and the anode to the cathode, in that, if the grid is close to the cathode, but the anode is relatively remote, the effectiveness of the grid in determining the anode current will be much greater, and will therefore give a higher value of u than if the anode is closer to the grid and cathode. Unfortunately, there is a snag in that the anode current resistance of the tube, Ra, is also related to the anode/cathode spacing, and becomes higher as the anode/cathode spacing is increased. The closeness of the pitch of the wire spiral which forms the grid also affects the anode current resistance in that a close spacing will lead to a high Ra, and vice versa.
The stage gain (M) of a simple tube amplifier, of the kind shown in FIG. 4, is given by the equation
M = μR/(R + Ra)
...so that a low impedance tube, such as a 6SN7 (typical I_a = 9mA, R_a = 7.7k, μ = 20), which has close anode-grid and grid-cathode spacings, and a relatively open pitch in the grid wire spiral, will have a high possible anode current but a low amplification factor, while a high impedance tube such as a 6SL7 (typical I_a = 2.3mA, R a = 44k, μ = 70) will have a low stage gain unless the circuit used has a high value of anode load resistance (R), and this, in turn, will demand a high value of HT voltage.
FIG. 3 Triode tube characteristics
FIG. 4 Simple tube amplifier
The space charge
Although a cloud of electrons will surround any heated cathode mounted in a vacuum, and will act as a reservoir of electrons when these are drawn off as anode current, their presence becomes of particular importance when a negatively charged control grid is introduced into the system, in that the electron cloud will effectively fill the space between the cathode and the grid, and will act as the principal source of electrons.
The presence of this electron cloud- known as the space charge - has several important operational advantages. Of these the first is that, by acting as an electron reservoir, it allows larger, brief duration, current flows than would be available from the cathode on its own, and that it acts as a measure of protection to the cathode against the impacts of positive ions created by electronic collisions with the residual gases in the envelope, since these ions will be attracted towards the more negatively charged cathode. Finally, left to itself, the electronic emission from the cathode suffers from both 'shot' and 'flicker' noise, a current fluctuation which is averaged out if the anode current is drawn from the space charge.
This random emission of electrons from a space charge depleted cathode is used to advantage in a 'noise diode', a wide-band noise source which consists of a tube in which the cathode is deliberately operated at a low temperature to prevent a space charge from forming, so that a resultant noisy current can be drawn off by the anode.
In the case of a triode used as an output power tube, where large anode currents are needed, the grid mesh must be coarse, and the grid-cathode spacing must be close.
This limits the formation of an adequate space charge in the grid-cathode gap, and, in its absence, the cathode must have a higher emission efficiency than would be practicable with an indirectly heated system, and this means that a directly heated filament must be used instead. Usually, the filament voltage will be low to minimize cathode induced 'hum', and the filament current will be high, because of the size of the filament (2.5A at 2.5V in the case of the 2A3 tube). Directly heated cathodes are also commonly used in tube HT rectifiers, such as the 5U4 or the 5Y3, because the higher cathode emission reduces the voltage drop across the tube and increases the available HT output voltage by comparison with a similar power supply using an indirectly heated cathode type.
Tetrodes and pentodes
Although the triode tube has a number of advantages as an amplifier- such as a low noise and low distortion factor- it suffers from the snag that there will be a significant capacitance, typically of the order of 2.5pF, between the grid and the anode. In itself, this latter capacitance would seem to be too small to be troublesome, but, in an amplifying stage with a gain of, say, 100, the Miller effect will increase the capacitance by a factor of 101, increasing the effective input capacitance to 252.5pF, which could influence the performance of the stage.
When triode tubes were used as RF amplifiers, in the early years of radio, this anode-grid capacitance caused unwanted RF instability, and the solution which was adopted was the introduction of a 'screening' grid between the triode control grid and its anode, which reduced this anode-grid capacitance, in the case of a screened grid or tetrode tube, to some 0.025pF.
A further effect that the inclusion of a screening grid had upon the tube characteristics was to make the anode current, in its linear region, almost independent of the anode voltage, which led to very high values for R a and la. Unfortunately, the presence of this grid caused a problem that when the anode voltage fell, during dynamic conditions, to less than that of the screening grid, electrons hitting the anode could cause secondary electrons to be ejected from its surface - especially if the anode was hot or its surface had been contaminated by cathode material - and these would be collected by the screening grid, which would cause a kink in the anode current/voltage characteristics.
While this might not matter much in an RF amplifier, it would cause an unacceptable level of distortion if used in an audio amplifier stage.
Two solutions were found for this problem, of which the simplest was to interpose an additional, open mesh, grid between the anode and the screening grid. This grid will normally be connected to the cathode, either externally or within the tube envelope, and is called the suppressor grid because it acts to suppress the emission of secondary electrons from the anode.
Since this type of tube had five electrodes it was called a 'pentode'. A typical small- signal pentode designed specifically for use in audio systems is the EF86, in which steps have also been taken to reduce the problem of microphony when the tube is used in the early stages of an amplifying system. The EF86 also has a wire mesh screen inside the glass envelope, and surrounding the whole of the electrode structure.
This is connected to pins 2 and 7, and is intended to lessen the influence of external voltage fields on the electron flow between the tube electrodes.
In use, a small-signal pentode amplifying stage will give a much higher stage gain than a medium impedance triode tube (250 in comparison with, say, 30). It will also have a better HF gain due to its lower effective anode-grid capacitance. However, a triode gain stage will probably have a distortion figure, other things being equal, which is about half that of a pentode.
The second solution to the problem of anode current non-linearity in tetrodes, particularly suited to the output stages of audio amplifiers, was the alignment of the wires of the control grid and screening grid so that they constrained the electron flow into a series of beams, which served to sweep any secondary electrons back towards the anode- a process which was helped by the inclusion within the anode box of a pair of 'beam confining electrodes', which modified the internal electrostatic field pattern.
These are internally connected to the cathode, and take the form shown in FIG. 5.
These tubes were called beam-tetrodes or kinkless tetrodes, and had a lower distortion than output pentodes. Tubes of this type, such as the 6L6, the 807, the KT66 and KT88, were widely employed in the output stages of the high quality audio amplifiers of the 1950s and early 1960s.
Both pentodes and beam-tetrodes can be used with their screen grids connected to their anodes. In this mode their characteristics will resemble a triode having a similar grid-cathode and grid-anode spacings to the grid-cathode and grid-screen grid spacings of the pentode. The most common use of this form of connection is in power output stages, where a triode connected beam tetrode will behave much like a power triode, without the need for a directly heated (and hum-inducing) cathode.
FIG. 5 Construction of beam-tetrode (courtesy, RCA)
Tube parameters
In addition to the anode current resistance, R a, and the amplification factor, p, mentioned above, there is also the tube slope or mutual conductance (gm) which is a measure of the extent to which the anode current will be changed by a change in grid voltage. Traditionally this would be quoted in milliamperes per volt (mA/V or milli- Siemens, written as mS), and would be a useful indication of the likely stage gain which would be given by the tube in an amplifying circuit.
This would be particularly helpful in the case of a pentode amplifying stage, where the value of R a would probably be very high in comparison with the likely value of load resistance. ( For example, in the case of the EF86, R a is quoted as 2.5M-Ohm and the gm is 2mA/V.) In this case, the stage gain (M) can be determined, approximately, by the relationship M ---g_m.R_L, which, for a 100k anode load would be ~-200". The various tube characteristics are defined mathematically as,
R a = dVa/dI_a t a constant grid voltage,
gm = -dI_a/dVg at a constant anode voltage, and
u = -dVa/dVg at a constant anode current.
In these equations the negative sign takes account of the phase inversion of the signal.
These parameters are related to one another by the further equation,
gm = u / R_a or u—g_m R_a
Gettering
The preservation of a high vacuum within its envelope is essential to the life expectancy and proper operation of the tube. However, it’s difficult to remove all traces of residual gas on the initial pumping out of the envelope, quite apart from the small but continuing gas evolution from the cathode, or any other electrodes which may become hot in use. The solution to this problem is the inclusion of a small container, known as a boat, mounted somewhere within the envelope, but facing away from the tube electrodes, which contains a small quantity of reactive material, such as metallic calcium and magnesium.
The boat is positioned so that after the pumping out of the envelope has been completed, and the tube had been sealed off, the getter could be caused to evaporate on to the inner face of the envelope by heating the boat with an induction heating coil.
Care is taken to ensure that as little as possible of the getter material finds its way on to the inner faces of the tube electrodes, where it may cause secondary emission, or on to the mica spacers where it may cause leakage currents between the electrodes.
While this technique is reasonably effective in cleaning up the gas traces which arise during the use of the tube, the vacuum is never absolute, and evidence of the residual gas can sometimes be seen as a faint, deep blue glow in the space within the anode envelope of a power output tube. If, however, there is a crack in the glass envelope, or some other cause of significant air leakage into the tube interior, this will become apparent because of a whitening of the edges of the normally dark, mirror-like surface of the getter deposit on the inside of the tube envelope. A further sign of the ingress of air into the tube envelope is the presence of a pinkish-violet glow which extends beyond the confines of the anode box. By this time the tube must be removed and discarded, to prevent damage to other circuit components through an increasing and uncontrolled current flow.
Cathode and heater ratings
For optimum performance, the cathode temperature should be maintained, when in use, at its optimum value, and this requires that the heater or filament voltages should be set at the correct levels. Since the voltage of the domestic AC power supply is not constant, the design ratings for the heater or filament supply must take account of this.
However, this is not as difficult to do as it might appear. For example, Brimar, a well-known tube manufacturer, make the following recommendations in their Valve and Teletube Manual: 'the heater supply voltages should be within +5% of the rated value when the heater transformer is fed with its nominal input voltage, provided that the mains power supply is within +10% of its declared value.
An additional requirement is that, because of inevitable cathode-heater leakage currents, the voltages between these electrodes should be kept as low as possible, and should not exceed 200V. Moreover, there must always be a resistive path, not exceeding 250k.Q, between the cathode and heater circuits. (As can be seen from some of the audio amplifier circuits shown in Section 6, electronic circuit designers often neglected this particular piece of advice.) As a practical point, the wiring of the heater circuit, which is usually operated at 6.3V AC, will normally be installed as a twisted pair to minimize the induction of mains hum into sensitive parts of the system, as will the heater wiring inside the cathode tube of low noise tubes, such as the EF86. With modem components, such as silicon diodes and low-cost regulator ICs, there is no good reason why the heater supplies to high quality tube amplifiers should not be derived from smoothed and stabilized DC sources.
It has been suggested that the cathodes of tubes can be damaged by reverse direction ionic bombardment if the HT voltage is applied before the cathode has had a chance to warm up and form a space charge, and that the tube heaters should be left on to avoid this problem. In practice, this problem does not arise because gaseous ions are only formed by collisions between residual gas molecules and the electrons in the anode current stream. If the cathode has not reached operating temperature there will be little or no anode current, and, consequently, no gaseous ions produced as a result of it. Brimar specifically warn against leaving the cathode heated, in the absence of anode current, in that this may lead to cathode poisoning, because of chemical reactions occurring between the exposed reactive metal of the cathode surface and any gaseous contaminants present within the envelope. Unfortunately, the loss of electron emissivity as the cathode temperature is reduced occurs more rapidly than the reduction in the chemical reactivity of the cathode metals.
Indirectly heated HT rectifier tubes have been used, in spite of their lower operating efficiency, to ensure that the full HT voltage was not applied to the equipment before the other tubes had warmed up, but this was done to avoid the HT rail over-voltage surge which would otherwise occur, and allow the safe use of lower working voltage, and less expensive, components such as HT reservoir, smoothing, or inter-tube coupling capacitors.
Microphonics
Any physical vibration of the grid (or filament, in the case of a directly heated cathode) will, by altering the grid-cathode spacing, cause a fluctuation of the anode current, and this will cause an audible tinging sound when the envelope is tapped--an effect known as microphony in the case of a tube used in audio circuitry. Great care must therefore be taken in the manufacture of tubes to maintain the firmness of the mounting of the grids and other electrodes. This is done by the use of rigid supporting struts whose ends are located in holes punched in stiff mica disc shaped spacers which, in turn, are a tight fit within the tube envelope.
Since a microphonic tube will pick up vibration from any sound source, such as a loudspeaker system in proximity to it, and convert these sounds into (inevitably distorted) electrical signals which will be added to the amplifier output, this can be a significant, but unsuspected, source of signal distortion which won’t be revealed during laboratory testing on a resistive dummy load. Since it’s difficult to avoid tube microphony completely, and it’s equally difficult to sound proof amplifiers, this type of distortion will always occur unless such tube amplifier systems are operated at a low volume level or the amplifier is located in a room remote from the loudspeakers.
Solid State Devices
Bipolar junction transistors
'N'- and 'P'- type materials
Most materials can be grouped in one or other of three classes, insulators, semiconductors or conductors, depending on the ease or difficulty with which electrons can pass through them. In insulators, all of the electrons associated with the atomic structure will be firmly bound in the valency bands of the material, while in good, usually metallic, conductors many of the atomic electrons will only be loosely bound, and will be free to move within the body of the material.
In semiconductors, at temperatures above absolute zero (0K or-273.15C) electrons will exist in both the valency levels where they are not free to leave the atoms with which they are associated, and in the conduction band, in which they are free to travel within the body of the material. This characteristic is greatly influenced by the 'doping' of the material, which is normally done, during the manufacture of the semiconductor material, by introducing carefully controlled amounts of specific impurities into the molten mass from which the single semiconductor crystal is grown.
The most common semiconductor material in normal use is silicon, because it’s cheap and readily available, and has good thermal properties. Germanium, the material from which all early transistors were made, has electrical characteristics which are greatly influenced by its temperature, which is inconvenient in use. Also it does not lend itself at all well to contemporary mass-production techniques.
In the case of silicon, which has very little conductivity in its undoped 'intrinsic' form, the most common dopants are boron or aluminum which give rise to a semiconductor with a deficiency of valency electrons, usually referred to as holes- called a 'P'- type material- or phosphorus, which will cause the silicon to have a surplus of valency electrons, which forces some of them into the conduction band. Such a semiconductor material would be termed 'N' type. Both P-type and N-type silicon can be quite highly conductive, depending on the doping levels used.
Fermi levels
The electron energy distribution in single-crystal P- and N- type materials is shown in FIG. 6, and the mean electron energy levels, known as the Fermi levels, are shown as dotted lines on the diagrams. If two such differently doped semiconductor materials are in intimate contact with one another, electrons will diffuse across the junction so that the Fermi level is the same on both sides. This will cause a shift in the relative electrical potential of the two doped regions as is shown in FIG. 7, and a number of interesting effects arise from this.
FIG. 6 Average electron energy levels in semi-conductors as a function of
temperature or doping level
FIG. 7 Mechanism by which potential barrier arises at P-N junction
(In practice, the only way by which such an intimate contact could be achieved within a single crystal of a semiconductor material is by diffusing, say, a P-type impurity into a wafer of N-type material, so that where these doped regions came into contact there would be a predominance of P-type silicon on one side of the boundary and of N-type silicon on the other.)
The P-N potential barrier and the depletion zone
The first of the phenomena which occur at the P-N junction is that the electron conduction bands are displaced, so that electrical current (which consists, in the philosophy of this guide, of a movement of electrons from one place to another) cannot flow through a semiconductor junction, even in the forward direction, from the electron rich N-type zone into the electron deficient P-type one, until a high enough potential exists to overcome the potential barrier (the voltage gap between the two conduction bands). This is about 0.58V in a silicon junction at room temperature, and decreases as the junction temperature is increased.
The second effect, shown graphically in FIG. 8, is the creation of a depletion zone, in which the diffusion of electrons across the junction from the electron rich N region into the electron deficient P region fills in the notional holes (labeled with '+' signs in the drawing) where electrons should have been but were not. This leaves a region where there are no electrons at all, and current won’t flow through it. The effect on the width of the depletion zone of the voltage across the junction is shown, schematically, in FIG. 9.
FIG. 8 Growth of depletion layer on either side of a P-N junction. a. Reverse
bias wide depletion band, P N; c. Small forward bias. narrow depletion band.
no current ~ Little or no current. Anode (~ Cathode; b. No bias, normal depletion
band, no current; d. Large forward bias, no depletion band, large current.
The effect of the voltage across a semiconductor junction, in this case silicon, on its current flow is illustrated in FIG. 10. When a forward (i.e. conduction direction) voltage is applied across such a P-N junction, little current will flow until the voltage exceeds the potential barrier, when the current flow will increase rapidly until any further increase is limited by the conducting resistance of the semiconductor material or of the external circuit.
In the reverse or non-conducting direction there will always be a small leakage current which will gradually increase as the reverse potential is increased, up to the voltage at which the junction breaks down. This breakdown is due to what is termed an avalanche effect, in which the electrons which form the reverse leakage current, accelerated by the reverse potential, reach a high enough kinetic energy to cause ionization in the junction material. This ionization releases further electrons which, in turn, will be involved in further collisions, increasing yet further the leakage current.
In general, such reverse breakdown effects are to be avoided because the heat involved causes a temperature rise at the junction which can permanently damage the diode.
However, some diodes are deliberately manufactured, by the use of very high doping levels, so that reverse breakdown occurs at voltages which are low enough for thermal damage to be avoided. These are commonly called Zener diodes, though this term is really only appropriate for diodes having a reverse breakdown voltage below about 5.5V.
FIG. 9 Influence of external potential on width of depletion zone in junction 'diode'.
The semiconductor diode
FIG. 10 Normal forward and reverse characteristics of silicon and germanium
junction diodes
Junction capacitance
An additional feature of the semiconductor junction is its capacitance. If the junction is reverse biased this capacitance is relatively small, and has the characteristics shown, approximately, by the relationship
C=k / √ V
where V is the voltage across the junction, and k is some constant which depends on the doping level of the material, and the area and temperature of the junction. The capacitance of a forward biased junction is much larger, and much more strongly dependent on the doping level, the junction temperature and the forward current.
Typical values of capacitance for a reverse biased small-signal diode will be of the order of 2-15pF, while in the case of a forward biased junction the capacitance could well be some ten times this value, depending on the diode current. (In the case of transistors, where the relationship is much more complicated, the manufacturers seldom attempt to specify the input (base-emitter) capacitance, but lump this factor, along with other things which affect the transistor HF performance, in a general term, called the HF transition frequency (fT) which is the frequency at which the transistor current gain has fallen to unity in value.) Transistor action
If the simple P-N semiconductor junction is elaborated so that it has three layers, For example to make an NPN structure, of the kind shown schematically in FIG. 11, this can be made to act as an amplifying device, known as a junction transistor. In practice, one of the N-type layers will be only lightly doped, and this region, because of its function, will be called the collector. The other N-type region will be heavily doped, to reduce its conducting resistance, and since it’s the source of the current flow it’s termed the emitter. The doping levels chosen by the manufacturer for the intermediate P-type region, known as the base, will mainly be determined by the intended use of the device.
FIG. 11 Single crystal junction transistor structures
If a reverse voltage is applied across the collector-base junction, no significant current will flow so long as the applied voltage is less than the breakdown voltage of the junction. If a small forward voltage is applied across the base-emitter junction, once again no current will flow until the forward voltage reaches the potential barrier level.
However, the differences in doping types and the existence of potential differences across the junction will cause depletion zones to form, within the material, on either side of the base-emitter and base-collector junctions, so that, if the base layer is thin, the depletion zone may extend throughout its entire thickness.
Referring to the schematic drawing of FIG. 11, if the base is made more positive, electrons will flow from the emitter into the base region, but they won’t necessarily flow into the base circuit since, depending on the geometry of the junctions and the positive potential on the collector, a large proportion of the electrons which set out from the emitter will be swept straight through the base region and into the collector.
This leads to a base voltage vs. collector current relationship of the kind shown in FIG. 12, and allows the device to control a relatively large collector current by a relatively low base voltage.
FIG. 12 Relationship between collector current and collector voltage for
small signal silicon transistor
In this respect, the bipolar junction transistor (BJT) behaves in a similar manner to the thermionic tube, except that the grid of a tube is essentially an open-circuit electrode, whereas the base region of the junction transistor has a relatively low resistance path- whose value depends on the emitter current - to the emitter. Also, while in a tube the anode current is controlled by its grid voltage, in a junction transistor it’s more correct to regard the collector or emitter current as being controlled by the current flowing in the base circuit, which, as shown in FIG. 10, has a very non-linear relationship with the applied voltage.
Junction transistors are available in a range of packages, suitable for a very wide range of collector voltages, currents and power dissipations. Provided that their dissipations are not exceeded and the integrity of the encapsulation has not failed, all modem semiconductor devices have a virtually indefinite life.
The mutual conductance of a bipolar junction transistor is very high, and can be determined from the relationship
gm= Ic(q/kT)
where q is the charge on the electron (1.602 x 10^-19), k is Boltzmann's constant, and T is the absolute temperature.
This means that the mutual conductance is dependent principally upon the collector current, and for a junction temperature of 25C the slope will be about 39S/A. For a small-signal transistor, at 1mA, a practical value of gm could be 40mS (mA/V). This allows high stage gains with relatively low collector load resistances.
PNP and NPN transistors
A major practical advantage of the junction transistor is that, because P-type and N- type materials can be made with equal facility, it’s possible to make PNP (negative collector voltage) transistors almost as easily as NPN (positive collector rail voltage) ones, by the simple substitution of differently doped materials. The availability of these two types of device is of great convenience to the innovative circuit designer.
The schematic symbols used for these, and other discrete semiconductor devices, are shown in FIG. 13.
Junction FETs Depletion MOSFETs Junction transistors Enhancement MOSFETs N-channel P-channel
FIG. 13 Transistor circuit symbols
It should be understood, however, that these complementary types of transistor are not identical, since there is a basic difference in the way in which current will pass through N- and P-type materials. In an N-type region, where there is a surplus of electrons, current flow is due to the movement of these electrons, exactly as in a metallic conductor. On the other hand, in a P-type material, the shortage of electrons results in vacancies occurring in the valency bands in places where electrons should normally be.
In the semiconductor engineer's parlance, these absent electrons are referred to as holes, and behave in much the same way as positively charged electrons. The difference, though, is in their speed of flow, because the apparent movement of a hole occurs when it’s filled by an electron which has moved, in response to some electrical field, and has left a hole where it had been. This hole will, in turn, be filled by another electron which will leave yet another hole, and so on, like marbles on a solitaire board.
The effect of this is that current flow in a P-type region is both slower and more noisy than in an N-type material.
A practical consideration which follows from this difference is that small-signal PNP transistors, which have an N-type base region, have a slightly lower noise factor than their equivalent NPN versions, though, in modem devices, this difference is slight. On the other hand, the greater mobility of the electrons in the relatively broad emitter and collector regions means that the HF performance of an NPN transistor will be better, other things being equal, than a PNP one. This is a factor which should be remembered in relation to power output transistors.
Junction field effect transistors
It has been seen that the presence of a reverse potential across a semiconductor junction will increase the width of the depletion zone on either side of the junction, and vice versa. Field effect transistors, normally called FETs, make use of this effect to allow a voltage applied to a reverse biased P-N junction to control the flow of current along a strip of suitably doped semiconductor material, called the channel, from an input connection - called the source - at one end of the channel to an output connection- called the drain- at the other end, as shown in FIG. 14a. This control electrode is called the gate, and behaves like the control grid in a tube. Usually there will be two such parallel connected gate regions diffused into either side of the channel to increase the effectiveness of this electrical control action.
FIG. 14a Schematic construction of N- and P-channel junction FET
FIG. 14b Effect of pinch off in junction FET
If there is a voltage difference between the two ends of the channel, as will normally be the case, this will have the effect of making the depletion zone somewhat unsymmetrical, as shown in FIG. 14b. As the drain voltage is increased, in respect to the gate, the two depletion zones will move closer together until the channel is pinched off entirely, and current flow through this depleted region, still controlled by the gate voltage, is due to quantum-mechanical tunneling. In this condition, which is that of normal operation, the drain current is almost completely independent of the drain voltage, as shown in FIG. 15. The relationship between gate voltage and drain current is shown in FIG. 16, and is much more linear than for a bipolar junction transistor.
FIG. 15 Junction FET drain current characteristics at voltages above pinch
off
FIG. 16 Conduction characteristics of typical junction FET
Breakdown potentials and operating characteristics
In use, the junction FET behaves very much like a thermionic tube, in that the gate is effectively an open-circuit electrode so long as the gate-source potential is reverse biased. If it becomes forward biased the gate-channel junction will be effectively the same as a forward biased junction diode. It’s commonly supposed that all FETs are sensitive to, and can be destroyed by, the high voltages associated with inadvertent electrostatic discharges, but junction FETs are, in reality, no more fragile in this respect than any other small-signal diode.
By comparison with the bipolar junction transistor, FETs are relatively low voltage, low power devices, with maximum drain-gate voltages typically restricted to about 25--40V, and free-air dissipations of about 400mW. Although normal mutual conductance figures lie in the range of 2-10mS (mA/V), the in-circuit stage gain is limited by the fact that the low drain-gate breakdown voltage precludes the use of high drain load resistance values, unless very small drain currents are acceptable. Once again, the stage gain (M) can be calculated, approximately, from the equation M = gmRL
The input and drain-gate capacitances of an FET are similar, at about 2pF, to those of a small power triode tube. Neither FETs nor junction transistors are at all microphonic unless they are mounted at the end of long, flexible leads in proximity to some charged body. Modern junction FETs are capable of very low voltage noise figures, except that their high input impedances (typically some 10^12 ohms) mean that a high impedance input circuit can generate high values of Johnson (thermal) noise.
Insulated gate field effect transistors These devices, known as IGFETs or MOSFETs (metal oxide/silicon field effect transistors) have become the most widely used, in the greatest range of styles, of all the field effect transistors. Their method of operation is exceedingly simple, and an attempt to make a device of this type was made by Shockley in the early 1950s. (His efforts were frustrated by his inability to obtain sufficiently pure semiconductor material.) A simple MOSFET is shown, schematically, in FIG. 17, and consists, in principle, of a strip of very lightly doped P-type material, into which, at either end, an N-type region has been diffused to allow electrical connections to be made to the strip. As in the case of the junction FET, these parts of the MOSFET are called the source, the drain and the channel. In manufacture the emitter contact pad (an area of vacuum deposited aluminum) is arranged to overlap both the source diffused zone and the 'P-' (a style of nomenclature which means lightly doped P-type) channel. This makes the device behave just like a very low gain NPN transistor with its base connected permanently to its emitter, a condition in which no collector current will flow.
FIG. 17 Simple insulated gate FET
If, however, a thin insulating layer (typically silica or silicon nitride in the case of a MOSFET made from silicon) is formed so that it covers the channel region, and a conducting layer (most commonly polycrystalline silicon) is formed on top of this insulating surface, then if a positive potential is applied to this conducting surface layer (also called the gate) it will induce a film of negative charges (electrons) across the gap between the source and the drain.
Since there is no difference between electrons produced by electrostatic induction and those due to any other cause, if a voltage is now applied between drain and source, a current will flow, and this current will be controlled by the gate voltage. This gate voltage vs. drain current relationship, shown in FIG. 18, can be made very linear (leading to a low distortion in an amplifying circuit) by suitable design of the MOSFET structure (see Siliconix Technical Article, TA 82-3). The style of MOSFET described above would be called an N-channel enhancement type, because the drain current, at zero gate voltage, would be exceedingly small, but would increase as the gate voltage was made more positive. Its method of construction is shown schematically in FIG. 19.
FIG. 18 Conduction characteristics of small signal MOSFET
FIG. 19 Construction of lateral P-channel enhancement MOSFET
There is, however, a further type of MOSFET called an N-channel depletion type, in which there is a significant current flow at zero gate-source voltage, but which will be reduced as the gate is made negative. The characteristics of this type of MOSFET are shown in FIG. 20, and its method of construction is shown in FIG. 21. In this, the device fabrication has been modified so that there is a thin conductive N-type layer connecting the source to the drain, underneath the gate electrode. When a positive voltage is applied between the drain and the source a current will flow, which will be reduced as the gate voltage is made more negative. In this respect the N-channel depletion MOSFET behaves in a manner which is almost identical to the thermionic triode, but without the tube's problems of microphony, fragility and loss of emission during use.
FIG. 20 Conduction characteristics of N-channel depletion MOSFET
FIG. 21 Depletion type MOSFET
A fundamental problem with the MOSFET, of whatever type, is that the channel will usually be long, and the channel conductivity will be low. In low power applications these characteristics won’t be particularly inconvenient, but in higher current or higher power usage they would be undesirable. Two methods are employed to lessen these difficulties, firstly to construct the MOSFET so that the channels are vertical, when the normal fabrication methods can make them short, and secondly to arrange the layout of the device so that there are a multiplicity of channels operating in parallel.
This method of construction allows very high values of mutual conductance to be obtained: 2-5S/A for a vertically diffused power MOSFET, as compared with 2-10mS for a small-signal lateral device, having an ID of, say, 5mA. Similarly, while channel resistances of 500-3k~ are normal for a small-signal MOSFET, on resistances (called RDS on) of a fraction of an ohm are common for power devices. The penalty incurred by this multiple-channel construction is that the gate-source capacitance is very high -- of the order of 1-2nF for an N-channel power MOSFET, and somewhat higher than this for an equivalent P-channel device. Because of the method of construction, the gate-drain capacitance will be smaller than this.
Because the electrostatic induction of a charge in the channel in response to the presence of an electrical potential on the gate is an almost instantaneous effect, MOSFETs have an excellent HF response, and this makes them prone to very high frequency oscillation (which may be too high to be seen on the monitor oscilloscope)
if the physical layout of the circuit in which they are used is poorly arranged. D-MOS and T-MOS devices normally have a somewhat lower F T than V-MOS or U-MOS forms, and the avoidance of parasitic oscillation is consequently rather easier to achieve in the D or T types.
Because of their very high values of F T, it’s easier to obtain low values of harmonic distortion and phase error by the use of overall negative feedback in power MOSFET based audio amplifiers than in similar circuits using bipolar junction transistors in their output stages. Also, in practice, power MOSFETs are more robust than similar bipolar junction transistors. These advantages may justify the use of the rather more costly power MOSFETs in high quality amplifier designs.
Since in RF amplifier use, the relatively high drain-gate capacitance of the MOSFET would lead to the same type of problem that exists in the triode tube, a solid state equivalent of the screened-grid tube, called a dual-gate MOSFET, is made using the type of construction shown in FIG. 22. These transistors are suitable for use as RF amplifiers, with the drain-gate #1 capacitance reduced to about 0.01 pF. Source + Substrate G1 G,, SiO-, Drain
FIG. 22 Lateral dual gate N-channel depletion MOSFET
Noise levels
Because junction transistors are fundamentally low input impedance devices, they can also offer very low levels of thermal noise, in good quality components in well-designed circuitry. Allowing for their higher input impedances, junction FETs are also available in very low noise versions. Small signal MOSFETs, however, tend to be relatively noisy, because of the mechanism by which the drain current is induced.
This is particularly true for P-channel MOSFETs because the induced current carrying zone is composed of holes, whose action in tumbling over one another when carrying current leads to higher values of flicker and shot noise than in the N-channel versions 4.
Integrated circuits (ICs)
These are multiple-component devices which can be made by any type, or combination of types, of semiconductor technology. In these, a range of active devices are combined with other passive components, within a single package, to produce a complete circuit module. In general these will only require the provision of input and output connections, and either a single, or a pair of, power supply rails, to carry out some specific electronic function. These were originally manufactured to carry out various logic functions in which the very small physical sizes of the devices would allow both high operating speeds and a large number of internally interconnected gates. However, this has become the area of the most rapid growth in the whole of the electronic component field, and ICs are now made for an enormous variety of applications, from digital microprocessors to low-noise, high-gain linear operational amplifiers (op amps). There are a number of incentives to the audio circuit designer to use linear ICs, such as op. amps, wherever their use would be satisfactory or appropriate. These are their high reliability (an IC containing a hundred transistors and a score of resistors or capacitors will be just as reliable, in general, as any single component of that type, and much more reliable than the same arrangement built up from individual resistors and transistors), their small size and their relatively low cost. In addition, since IC manufacture is an exceedingly competitive business, the internal circuitry used in the ICs will reflect the skills of some of the most gifted engineers in the field. In particular, op amps are now capable of performances, as general purpose audio gain blocks, which can hardly be matched, let alone bettered, by discrete component layouts.
Further reading Since the principal purpose of this guide is to explore the circuit designs employed in audio amplifiers, both tube based and solid-state, I have tried to restrict my analysis of semiconductor behavior to that needed by the reader to understand what is going on. If a more comprehensive explanation of the function and construction of solid-state devices is wanted, I would recommend the following sections in our Guide to Linear Electronics: Sect. 4; Sect. 5
===