An electronic amplifier is a device for increasing the power of a signal. It does this by taking power from a power supply and shaping the output to match the input signal. This process invariably introduces some noise and distortion into the signal, and the process cannot be 100% efficient - amplifiers will always produce some waste heat. An idealised amplifer can be said to be "a piece of wire with gain", the output is an exact replica of the input, only larger.

Different designs of amplifier are used for different types of applications and signals. We can broadly divide amplifiers into three categories - small signal amplifiers, low frequency power amplifiers and RF power amplifiers. Each of these calls for a slightly different design approach, mainly because of the physical limitations of the components used to implement the amplifier, and the efficiencies that can be realised.

Amplifiers can be implemented using transistors of various types, or vacuum tubes (valves). Other more exotic forms of amplifier are also possible using different types of device, but these will not be discussed in detail here to avoid complicating the picture too much. Such exotic amplifiers are often used for use for microwave or other extremely high frequency signals.

Amplifier circuits are classified as A, B, AB and C for analogue designs, and class D and E for digital or switching designs. For the analogue classes, each class defines what proportion of the input signal cycle is used to actually switch on the amplifying device:

  • Class A - 100% of the input signal is used (conduction angle a = 360 or 2π)
  • Class AB - more than 50% but less than 100% is used. (181 - 359, π < a < 2π)
  • Class B - 50% of the input signal is used (a = 180 or π)
  • Class C - less than 50% is used (0 - 179, a < π )

This can be most easily understood using the diagrams in each section below. For the sake of illustration, a bipolar junction transistor is shown as the amplifying device, but in practice this could be a MOSFET or vacuum tube device. In an analogue amplifier, the signal is applied to the input terminal of the device (base, gate or grid), and this causes a current to flow in proportion to the input between the output terminal and ground (collector, drain or anode). This current is obtained from the power supply. The voltage signal shown is thus a larger version of the input, but has been changed in sign (inverted) by the amplification. Other arrangements of amplifying device are possible, but that given (common emitter, common source or common cathode) is the easiest to understand and employ in practice. If the amplifying element is linear, then the output will be faithful copy of the input, only larger and inverted. In practice, transistors are not linear, and the output will only approximate the input. This is the origin of distortion within an amplifier. Which class of amplifier (A, B, AB or C) depends on how the amplifying device is biased - in the diagrams the bias circuits are omitted for clarity.

Table of contents
1 Class A
2 Class B and AB
3 Class C
4 Negative Feedback
5 A Practical Circuit
6 Class D and E
7 See Also

Class A

Class A amplifiers amplify over the whole of the input cycle. They are the usual means of implementing small-signal amplifiers. They are not very efficient - a theoretical maximum of 50% is obtainable, but for small signals, this waste of power is still extremely small, and can easily be tolerated. It is only when we need to create output powers with appreciable levels of voltage and current does Class A become problematic. In a Class A circuit, the amplifying element is biased such that the device is always conducting to some extent, and is operated over the most linear portion of its characteristic curve (known as its transfer function or transconductance curve). Because the device is always conducting, even if there is no input at all, power is wasted. This is the reason for its inefficiency.

If we wish to produce large output powers from a Class A circuit, the power wastage will become significant. For every watt delivered to the load, the amplifier itself will, at best, waste another watt. For large powers this will call for a large power supply and large heat sink to carry away the waste heat. Class A designs have largely been superseded for audio power amplifiers, though some audiophiles believe that Class A gives the best sound quality, due it being operated in as linear a manner as possible. In addition, aficionados prefer vacuum tube designs over transistors, for a number of reasons. One is that the characteristic curve of a valve means that distortion tends to be in the form of even harmonics, which sound more "musical" than odd harmonics. Another is that valves use many more electrons at once than a transistor, and so statistical effects lead to a "smoother" approximation of the true waveform - see shot noise for more on this. Field-effect transistors have similar characteristics to valves, so these are found more often in high quality amplifiers than bipolar transistors. Historically, valve amplifiers often used a Class A power amplifier simply because valves are large and expensive; the Class A design uses only a single device. Transistors are much cheaper, and so more elaborate designs that give greater efficiency but use more parts are still cost effective.

Class B and AB

Class B amplifiers only amplify half of the input wave cycle. As such they create a large amount of distortion, but their efficiency is greatly improved. This is because the amplifying element is switched off altogether half of the time, and so cannot dissipate power. A single Class B element is rarely found in practice, though it can be used in RF power amplifiers where the distortion is unimportant. However Class C is more commonly used for this.

A practical circuit using Class B elements is the complementary pair or "push-pull" arrangement. here, complementary devices are used to each amplify the opposite halves of the input signal, which is then recombined at the output. This arrangement gives excellent efficiency, but can suffer from the drawback that there is a small glitch at the "joins" between the two halves of the signal. This is called crossover distortion. A solution to this is to bias the devices just on, rather than off altogether when they are not in use. This is called Class AB operation. Each device is operated in a non-linear region which is only linear over half the waveform, but still conducts a small amount on the other half. The result is that when the two halves are combined, the crossover is greatly minimised or eliminated altogether.

Class B or AB push-pull circuits are the most common form of design found in audio power amplifiers, and are sometimes used for RF linear amplifiers as well.

Class C

Class C amplifiers conduct over less than 50% of the input signal. As such the distortion at the output is gross, but very high efficiencies can be reached - up to 90% or so. Some applications can tolerate the distortion, such as audio bullhorns. A much more common application for Class C amplifiers is in RF transmitters, where the distortion can be vastly reduced by using tuned loads on the amplifier stage. The input signal is used to roughly switch the amplifying device on and off, which causes pulses of current to flow through a tuned circuit. The tuned circuit will only resonate at one particular frequency, and so the unwanted harmonics are suppressed, and the wanted full signal (sine wave) will be developed across the tuned load. Provided the transmitter is not required to operate over a very wide band of frequencies, this arrangement works extremely well. Any residual harmonics can be removed using a filter.

Negative Feedback

All amplifying devices are inherently non-linear, their physics dictates that they operate using a square law. While these devices can be treated as linear over some portion of their characteristic curve, the fact is that no device is truly linear. The result of non-linearity is distortion. As we have seen, the application dictates how much distortion we can tolerate. For hi-fi audio applications, instrumentation amplifiers and the like, distortion must be minimal. While careful design of each stage can achieve very good results, overall the best an open-ended (open loop) design can achieve is about 1% distortion. One way to reduce this further is to use negative feedback. This involves feeding a portion of the output back to the input in such a way that it cancels out part of the input. Naturally the main effect of this is to reduce the overall gain of the system, which might seem counter productive. However, all of the unwanted signals that are introduced by the amplifier are also fed back, and, since they are not part of the original input, become added to the input, but in opposite phase. The result is that the system as a whole becomes totally linear, because the input now "anticipates" all the distortions that will arise. With feedback, distortion can be lowered to negligible levels - 0.001% being typical. By the same means, noise is also reduced. Even effects such as crossover distortion can be eliminated using negative feedback. With feedback, the amplifier itself can change over time (due to deterioration of its components, changes in temperature, etc), with absolutely no change in its performance. While feedback would appear to be a universal fix for all the problems an amplifier can suffer from, there are many who believe that negative feedback is a bad thing.

The arguments against feedback include the fact that being a loop it takes a finite time to react to an input signal, and for this short period the amplifier is "out of control". A musical transient that is of the same order as this period will be grossly distorted, even though the amplifier will show incredibly good distortion performance on steady-state signals. Proponents of feedback refute this, saying that the feedback "delay" is of such a short order that it represents a frequency vastly outside the bandwidth of the system, and such effects are not only inaudible, but not even present, as the amplifier will not respond to such high frequency signals. The argument has been the source of controversy for many years, and has led to all sorts of interesting designs to deal with it - such as feedforward amplifiers. The fact remains that the majority of modern amplifiers use considerable amounts of feedback, though the best audiophile designs seek to minimise this as much as possible.

The concept of feedback is used in operational amplifiers to precisely define gain, bandwidth and other parameters.

A Practical Circuit

For the purposes of illustration, this practical amplifier circuit is described. It could be the basis for a moderate audio power amplifier. It features a typical design found in modern amplifiers, with a class AB push-pull output stage, and uses some overall negative feedback. Bipolar transistors are shown, but this design would also be realisable with FETs or valves.

The input signal is coupled through capacitor C1 to the base of transistor Q1. The capacitor allows the AC signal to pass, but blocks the DC bias voltage established by resistors R1 and R2 so that any preceding circuit is not affected by it. Q1 and Q2 form a differential amplifier, in an arrangement known as a long-tailed pair. This arrangement is used to conveniently allow the use of negative feedback, which is fed from the output to Q2 via R7 and R8. The amplified signal from Q1 is directly fed to the second stage, Q3, which provides further amplification of the signal, and the d.c. bias for the output stages, Q4 and Q5. R6 provides the load for Q3. So far, all of the amplifier is operating in Class A. The output pair are arranged in Class AB push-pull, also called a complementary pair. They provide the majority of the current amplification and directly drive the load, connected via d.c. blocking capacitor C2. The diodes D1 and D2 provide a small amount of constant voltage bias for the output pair, just biasing them into the conducting state so that crossover distortion is minimised. This design is simple, but a good basis for a practical design because it automatically stabilises its operating point, since feedback internally operates from d.c up through the audio range and beyond. Further circuit elements would probably be found in a real design that would roll off the frequency response above the needed range to prevent the possibility of unwanted oscillation. Also, the use of fixed diode bias as shown here can cause problems if the diodes are not both electrically and thermally matched to the output transistors - if the output transistors turn on too much, they can easily overheat and destroy themselves, as the full current from the power supply is not limited at this stage. Calculating the values of the resistors is left as an exercise for the reader.

Class D and E

Class D, "digital" amplifiers use several, usually a binary number, of high power "switches" (i.e. transistors usually). The input signal must be "sampled" more than twice as often as the highest frequency of interest. The samples are turned into numbers. The numbers are used to switch combinations of the switches off and on, and the current from these is fed through a summing electronic mixer , usually a network of resistors . Since any individual transistor is either on or off, the amplifer is efficient.

A badly designed digital amplifier can be a multiplying electronic mixer. It can add difference-frequencies between the sample rate and the desired signals. These show up as unwanted low frequency noise. In order to avoid these, an electronic filter must be placed on the input to throw away all signals with frequencies more than half that of the sampling frequency. Nyquist proved that one must have at least one sample for the high part of the wave, and one for the low part of the wave. This sample rate, two times the maximum interesting frequency, is called the "Nyquist" frequency.

Another problem with class D amplifiers is that the sets of switches can't perfectly match what the output should be. One can get closer with more switches and resistors. This error is called "quantization error" and appears as harmonic distortion. In practice, most people can't hear quantization error in speech when the range of numbers is wider than about one in 4000, and the sampling rate is greater than 8,000 times per second. This is the rate and width that long distance telephone lines use.

Class E amplifiers use pulse width modulation to amplify. The plan here is that the output signal is turned off and then on periodically. The time off and on is made to be proportional to the desired output amplitude. Since this amplifer just has one switch, it can be cheaper than any other class.

However, it also has more types of potential error than any other class. Class E amplifiers take samples, and thus have a Nyquist frequency and require a filtered input. It's also common to use digital circuits to control the duty cycle by counting some very fast periodic clock signal. When this is done, class E amplifiers also have quantization error. Last but not least, class E amplifiers have sharp edges, so they can have severe harmonic distortion though it occurs at frequencies above the frequencies of interest.

Class E amplifers are used to control motors. In fact, it's hard to find any other type of motor controller for small DC motors. They have also been used as transmitters for commercial AM and shortwave radio, although more modern systems use class D amplifiers, which introduce lower distortion.

See Also