The Layperson’s Guide to Amplifiers

Al Sekela, 4 July 2005

Introduction

This document describes high-end audio amplifiers with minimum use of mathematics to assist those without technical training in selecting the best type for their needs.  Some technical definitions are necessary.  Amplifiers have been developed along several lines over many years, and it is not possible to understand the reasons for the diversity without knowing what the jargon means.

Definitions

Ohm’s Law: V = I×R

This basic relationship expresses the ideal case where a DC voltage V depends linearly on the current “I” (measured in amperes) through something that has resistance “R” (measured in ohms).  When the voltage and current are time-varying, such as when they represent an audio signal, a similar relationship is V = I×Z.  In this case, “Z” is called the impedance and it may include resistance, capacitance, and inductance.  Capacitance and inductance relate to energy storage rather than dissipation as in the resistance, and will be discussed further as needed.

There are three impedances important for audio amplifiers: input, output, and load.  The input impedance is usually somewhere between 10,000 and 500,000 ohms, but in a few cases may be as low as 600 ohms.  It determines how much current the source (preamp, linestage, or CD player) has to deliver to drive the amplifier.  It is possible to cause the sound to suffer by loading a weak source with an amplifier with a lower input impedance.

The load impedance is the speaker impedance, usually 4 to 16 ohms.  It is important to be sure the speaker impedance is not lower than the amplifier’s rated load impedance.  This will be discussed later.

The output impedance is a theoretical thing.  There is no single object inside the amplifier called the output resistor.  It is the equivalent resistance within the amplifier caused by limitations in the output devices, the power supply, and the circuit arrangement, that prevents the amplifier from being an ideal voltage source.  It is usually very small, but in some cases may be a few ohms.

Power: P = V•I (measured in watts if voltage is in volts and current is in amperes)

In DC circuits, this is exactly true.  In AC circuits, some of the current may be used to store energy in capacitance and inductance, and does not represent real power.  To determine the amount of power when the AC signal is a pure sine wave, the expression is P= V•I•cosine{ø}, where the greek letter phi is commonly used to represent what is called the “phase angle” between the voltage and current waveforms.  There is a close relationship between sine and cosine functions and points on the circumference of a circle, so engineers commonly refer to phase angles as if they were angles between two radial lines drawn to different points of the circumference.  “Phase” has a more general use that we will encounter in the discussion of amplifier stability.

Power is related to impedance through Ohm’s Law.  Knowing the maximum voltage that an amplifier can produce defines the maximum power when the impedance is known.  This relates to safe operation, especially of transistor amps.

Gain

The factor by which the signal voltage is raised is called the “gain,” and is given in units of decibels, abbreviated “dB.” Most home audio products have line signal voltage maximums around 2 volts.  The power rating of the amplifier is directly determined by the power output into a rated load impedance.  40 volts of output gives 200 watts into 8 ohms (see the definitions of Ohm’s Law and the power relationship).  Such an amplifier has a voltage gain of 40/2, or 20 times.  This is 26 dB.

The mathematics of gain calculations involves logarithms.  This was established before computers, when most engineering calculations were done by hand and it was easier to look up logarithms and add and subtract them, than to multiply and divide numbers with lots of significant digits.  Decibels are logarithmic units, so components connected in series would have their gains stated in decibels added, rather than their gain factors multiplied, to get the overall gain.

Fourier Series (also known as harmonic analysis)

Interesting audio signals that represent speech and music are complex and generally nonrepetitive.  They were far too complicated to deal with directly in the early days of radio and electronics.  Engineers found that much progress could be made by treating specialized test signals that were repetitive.  They used a mathematical tool called the Fourier Series to break any repetitive signal down into a series of sine and cosine functions.  One requirement of the series is that only integer multiples of the fundamental frequency could be used in the additional sine and cosine functions.  These multiples are called the “harmonics” and numbered from the fundamental as number one.

An important consequence of Fourier Series analysis is that sharp corners in test or music signals require many harmonics to approximate faithfully.  This is why some amplifier tests involve “square” wave signals.  It would take an infinite number of harmonics to completely reproduce the sharp corners of an ideal square wave, so the rounding observed in real circuits is a quick way to display upper frequency limitations.

Bandwidth

The undamaged hearing of young humans is typically assumed to include the frequency range from 20 Hz to 20,000 Hz.  (“Hz” is the abbreviation for Hertz and is the accepted unit of frequency in cycles per second.) This means some humans can hear tones with fundamentals over this entire range.  Damage and age diminish the upper frequency acuity of humans, and there is some controversy over whether or not audio information above the tone response limit adds anything to perception of the reproduced sound.  In any case, the bandwidth of amplifiers is an important design criterion and is nearly always specified.  It is the frequency range over which some specified performance attribute, such as power or distortion, is true.  It determines how well the amplifier will reproduce both square wave test signals as well as real sounds with sharp transients.

Distortion

Imperfections in amplifier performance include changes to the shape of the signal waveform.  Ideally, the amplifier would only make the signal larger.  Changes to the shape are called distortion.  There are many kinds of distortion.  Harmonic distortion was the first to be measured and is often quoted in amplifier specifications as Total Harmonic Distortion, or THD, as a fraction of the undistorted signal.  This single number does not convey much information, as many studies have shown that humans typically do not notice even-integer harmonics as much as odd-integer harmonics, when these (Fourier Series) harmonics represent the distortion.

Bias

The active electronic devices in amplifiers are only linear over voltage and current ranges that are not close to zero.  To avoid excessive distortion from nonlinear behavior near zero voltage or current, the circuit is designed for a steady (DC) current or voltage to be applied during operation.  This is called the “bias.” Some tube amps require manual adjustment of bias current as the tubes age.  Note that this bias current does not appear in the speaker unless something is broken in the amplifier.

Amplifiers are classified as to whether the bias is such that the output devices always have current flowing (Class A), have current flowing through more than half of a sine-wave test signal (Class AB), or have current flowing for approximately one-half of a sine-wave test signal (Class B).  Class A uses the most standby power but is completely free of distortion caused by devices going back and forth between inactive (called “cutoff”) and active states during the music waveform.  There are other bias classes but these are not relevant to audio amplifiers.

Sometimes tube circuit bias schemes are further divided with suffixes 1 and 2.  Class A1, for example, means that no current flows into the control grid part of the tube at any time, while Class A2 means that control grid current may flow part of the time.  This distinction is important in the design of the driving ciruits that would have to supply the grid current.  The output tubes can supply more peak power in Class A2, but the driving circuits may then introduce distortion if not designed with care.

Feedback

Subtracting some of the output from the input is a trick that allows engineers to use lower-cost devices to get higher levels of performance.  This is called “negative feedback.” An amplifier can be made to oscillate, or generate a tone all by itself, if the output is added to the input rather than being subtracted.  This is called “positive feedback.”

Stability

An amplifier should not oscillate or generate its own tones.  If it does, it is considered to be unstable.  Such oscillation can burn out speakers and damage the amplifier.  If the amplifier is marginally stable, it may well avoid oscillation, but it will give a very unsatisfactory rendition of the signal.  Think of a public-address system where the microphone is in range of the speaker.  Feedback will cause it to squeal, but if the volume is reduced just enough so that it does not squeal, there will still be a hollow ringing quality to most of the speech.  Stability is related to the use of negative feedback as well as the impedance of the load.

Damping

Damping in general is the conversion of mechanical or electrical energy into heat.  What Americans call “shock absorbers” on cars are really damping devices.  Anything with both inertia and a restoring force (such as a car on springs or a woofer cone with its suspension) will oscillate when stimulated.  Adding damping to such an object causes the oscillations to die down when the stimulus is removed.  There can be too much as well as too little damping, depending on the circumstances.

In amplifier discussion, “damping factor” refers to the amplifier’s ability to absorb current developed within the speaker as the speaker attempts to oscillate.  The same word may be used to describe treatments to the amplifier components in order to reduce their tendency to vibrate.

Regulation

Regulation refers to a power supply’s ability to deliver a constant voltage in spite of load variations.  There are active circuits called regulators in some amps that maintain a constant voltage over a wide range of currents, but these are usually not used for the output stage, where the regulators would have to be as big as the audio part of the amp to avoid restricting it.  The natural regulation of the main supply is determined by how big it is.  Larger transformers and capacitors have less internal resistance and their output drops less as the load increases.

Ground

This topic generates more confusion than any other, even among professionals and manufacturers.  The safety ground is a physical rod or net buried in the earth and attached to the AC power service entrance.  The incoming neutral wire from the local step-down transformer and all the neutral conductors from all the circuits in the house are attached to this ground at one place, inside the breaker panel.  It is there to protect the house from lightning strikes to the utility wires in the neighborhood.

The ground wires inside the house’s power wiring are there to connect all grounded equipment together, so that a fault in one (say, a washing machine) will not apply line voltage to any person who happens to be in contact with it and another piece of equipment (say, a dryer).  That the safety circuit is also grounded is a further benefit, to prevent shock if, say, the faulty washing machine and a water pipe were touched at the same time.  A fault to the case of a properly-grounded piece of equipment will usually cause the fuse to blow or the circuit breaker to trip immediately.

Some audio equipment has two-prong plugs.  This gear is built to be inherently safe and free from electric shock hazard from the case (usually called “double-insulated”).  However, most large power amplifiers have three-prong plugs and must be connected to a properly grounded circuit to be safe.  Use of an adapter as a “cheater” opens the user to risk of lethal shock if something should fail inside the amplifier.

This safety ground has audio consequences.  If an audio system contains more than one piece of grounded equipment, there can be what is called a “ground loop” formed.  This takes place because the usual practice is to attach audio ground (the part of the circuit that is the reference point for the signal) to safety ground at one point inside each piece of equipment.

The ground loop consists of the safety ground connection between the pieces of equipment (usually through the power wiring inside the walls if different outlets are used) and the audio ground or negative lead in the interconnect cable that also connects them.  Hum and noise voltages induced in the safety ground wires inside the walls are added directly to any single-ended audio signal that passes between these pieces of equipment.  The hum can be very loud, or the degradation can be subtle.

If the ground loop is broken and the noise is reduced, it is strongly tempting to use an adaptor as a cheater to lift the safety ground, but this then requires the interconnect cable negative wire to carry the fault current, and is both illegal and dangerous.  Some gear may include the ability to separate the audio ground from the safety ground (so-called “ground-lift switch”).  Another approach would be to isolate the offending gear with audio transformers.  Adding separate dedicated power circuits for each piece of audio gear will reduce the ground loop noise as long as the circuits are installed correctly.  This may be the cheapest solution.

Ground loops may be tolerable if the AC power wiring does not couple much noise.  In these cases, having a good, low-impedance safety ground connection at the service entrance is important because some audio signal is coupled to the safety ground wiring from each piece, and the safety ground earth connection should not cause these different ghost signals to interact.  However, keep in mind the earth safety ground has nothing to do with reducing noise from other sources.

Attributes of Amplifiers

Different users have different goals for their audio systems, and hear things differently.  This is the basic reason for the large variety of amplifiers made for household use.  Another reason is that no one design is entirely free from drawbacks, and different designs favor different priority schemes of strengths.  The strengths and weaknesses relate to the natures of the components and materials available.  Audio is a very small market compared to the general consumer market, so it has always had to make do with components originally designed for other purposes.  To design a custom part is very expensive, and the cost to make only a few parts is prohibitive.

Users desire certain attributes in their systems.  It is worthwhile to discuss these before getting into the types of designs.

Power

The output power capacity of an amplifier is its basic attribute.  It determines the size, weight, power supply, and cooling requirements.  It is typically given over a stated bandwidth with an upper limit on the amount of harmonic distortion.  Any amplifier can be rated for more power if the amount of allowed distortion is increased.

Human hearing is logarithmic.  This means it takes ten times as much power to make something sound twice as loud.  Most music and dialog can be reproduced by most speakers with about a watt or two of amplifier power.  However, musical crescendos and gunshots typically require hundreds of watts to reproduce faithfully with the same speakers.  This is not a fault in the speakers, it is a simple consequence of the way humans respond to sound.

If the speakers are already selected, their power requirements in the listening room determine the minimum power rating needed in the amplifier.  Choosing the right power rating takes some care.  It is generally not a good idea to buy as much power capacity as possible, since some speakers can give good sound with modest amounts of power, and smaller amplifiers may have better sound.  However, see the section under “Clipping” below for discussion of the pitfalls of having too little power.

Some speakers use currents generated within their motors to provide braking force (damping) for proper bass performance.  These speakers require amplifiers with high current capabilities, even if they are not used to play at high levels, as these amplifiers are more capable of absorbing the braking currents without suffering from distortion.  Not all amplifiers with a given power rating will have the same current capability, so it is important to understand the speaker requirements and ask a lot of questions about the amplifiers if a high-current type is needed.

A related concept in audio amplifiers is the sense of authority or control.  Two different amps with similar power ratings may sound completely different in terms of authority.  The amp with more authority sounds more powerful even if it cannot deliver more maximum power.  This property is related to the power supply regulation: a bigger power supply has better regulation and the amp it serves has more authority.  It is also related to circuit tricks that make power supply regulation less critical.

Amplifiers are always rated for how much power they can produce into a given load impedance.  What is often not stated explicitly is whether the amplifier will be damaged if a lower load impedance is used.  It is safer to assume it will be unless the manufacturer states otherwise.

Clipping

When an amplifier is over-driven, the output cannot follow the shape of the input, but is instead truncated at a voltage related to the power supply voltage.  This truncation is called “clipping” because the tops of test sine waves are clipped off when viewed on an oscilloscope.  This is a dangerous situation because the shape change, usually to bass waveforms, introduces a lot of upper harmonics (consistent with Fourier analysis).  The energy associated with these harmonics is real, and goes through the speaker crossover into the tweeters.  Since these are usually not built to withstand very high sustained power levels, they can be burned out quickly.

Amplifiers with low rated output power, used with low-efficiency speakers, are the worst culprits.  The clipping may take place during transients that are in themselves amusical, so that the user is not aware of it until the tweeters stop playing altogether.

Different amps have different clipping characteristics.  Most tube amps are soft, while most transistor amps have hard clipping.  Hard clipping is closer to theoretical and is more dangerous for tweeters.

Heat

Different amplifier designs have different degrees of efficiency and standby mains power draw.  All the mains power not delivered to the speakers has to be dissipated as heat from the amplifier.  Very large audio power capacity may result in a lot of heat from the amplifier.  This has to be drawn efficiently into the room, as high temperatures are bad for electronic devices.  Air conditioning makes a lot of noise unless the installation is done with care and a lot of expense.  Therefore, it is important to think about how much heat the listening room can tolerate when selecting an amplifier.

It is a common misconception that tube amps generate a lot more heat than transistor amps.  For power levels upwards of 100 watts, the tube filament heat is a small part of the total heat.  Most of the heat comes from the nature of the circuitry (bias class).  Transistor amps that run cold at idle may suffer from lack of authority and have a harsh overall sound because their output stage bias is too low.  These amps run with bias Class AB that is closer to Class B than to Class A.  Amps that run closer to full Class A generate more heat.

Resolution

There is a limit to the smallness of features that can be resolved in a photograph, determined by pixel count if digital, or emulsion grain size if analog.  In similar fashion audio systems have resolution of quiet sounds and details of timbres limited by something in them.

Very few home systems can resolve all the information present on Redbook (16-bit, 44.1 KHz) audio compact discs.  This may seem surprising in an era when higher-resolution formats abound, but it reflects the chained weaknesses in sources, cables, electronics, and speakers, as well as room and supporting equipment interactions with the system.

Amplifiers in particular have fundamental resolution limits imposed by their designs, the natures of their components and materials, and how they are supported.  The resolution limits imposed by design and materials are seldom discussed, and it is difficult to find out these limits except by careful auditioning.  However, a graceful limit to amplifier resolution is important in creating the illusion of being transported to the musical performance space.

An important aspect of overall system resolution is electrical noise from the amplifier’s power supply.  If the system is fed from a single AC circuit, the noise from the amp’s power supply can get back into the source components and degrade their resolution.  An amplifier with a lot of power supply electrical noise may sound better on a dedicated AC circuit, so it is wise to audition amplifiers with the same kind of AC hookup that will be used in the system at home.

Another important system-related aspect of resolution is susceptibility to Radio Frequency (RF) noise.  Some amplifiers convert RF noise into spurious tones when the noise mixes with the signal.  The symptoms are a dry midrange and emphasized treble, as well as loss of resolution.  Since RF travels easily through small amounts of stray capacitance, it can get into the amplifier through the output terminals and power cord as well as the input cables.  This problem is not well recognized or understood, so it is difficult to evaluate different amplifier models except within the audio system.

Balanced versus Single-Ended

Some amplifiers are built with balanced (XLR) inputs.  For proper use, the impedances of the positive and negative input pins to audio ground should both be the same.  This is different from conventional, or single-ended, inputs, where the positive input impedance is high and the negative input resistance to audio ground is zero.  Remember, audio ground may not be the same as the external case safety ground, so it may not be possible to measure input resistances to the case.  The purpose of balanced inputs is to cancel out any external noise that couples equally to the positive and negative interconnect cable wires.

A related concept is differential amplification: only the voltage difference between the positive and negative pins is amplified.  This means the signal is present as both positive and negative copies throughout the amplifier.  It may mean that two sets of circuits are used, or that a clever circuit design accomplishes this.  The advantage is that any internal noise or external disturbance that affects the two copies of the signal equally is cancelled when they are converted into a speaker signal.  This property is called Common Mode Rejection.  Amplifiers with differential circuits usually use balanced inputs.

If done well, this advantage results in more resolution than is possible in single-ended amps of the same caliber.  The disadvantage is increased complexity and cost, and the requirement for balanced sources to drive the amplifier.  Note that the presence of XLR input terminals does not necessarily mean the input is balanced or that the circuitry is differential.  Some amplifiers have XLR inputs for convenience and convert the signal to single-ended internally.  This should be understood clearly, as some convenience XLR inputs do not present balanced loads and may degrade the sound from a balanced source.

Build Quality

Audio amplifiers are expensive and appearance counts.  However, some aspects of amplifier case design are sometimes taken to extremes that actually impair performance, while other aspects may be neglected.  It is difficult to find a high-end amplifier without a thick aluminum front panel these days.  Sometimes these are machined with stylish features.  They represent a problem, as the thick aluminum rings strongly, and the acoustic ringing can cause loss of resolution and introduction of tonal coloration.

The heaviest part in most amplifiers is the power transformer.  If this is not adequately supported, it can vibrate excessively from the speaker output and cause loss of bass resolution.  It can also vibrate mechanically from the AC power waveform.  Further, how the power transformer is oriented on the chassis affects how much hum and noise from its stray magnetic fields is coupled into the audio circuitry.  Listen carefully for hum or buzz from the amp itself and in the speakers with the inputs disconnected to see if there are problems with this.

Two different kinds of power transformers are toroidal (doughnut shaped), and E-I (suggested by the shapes of the core laminations) or conventional rectangular core.  Toroidal transformers have more power capacity for a given size but are able to transmit more noise from the AC line into the amplifier, and from the power supply circuits back onto the AC line.  This can be a serious concern if the audio system has to operate from a single AC circuit.  The relative inefficiency and limited bandwidth of the E-I design can be an advantage in these areas.  Some toroids are more prone to making a mechanical buzz from unbalanced AC (ie, if your neighbor is using his or her hair-dryer on the low setting, or some other heavy appliance is distorting the AC waveform).

Case panels may be undamped and able to introduce tonal coloration from their vibration modes.  The rear panel, in particular, usually has power and signal connectors attached.  Vibration in this panel may be especially damaging to resolution and tonal neutrality.

Types of Amplifiers

All audio amplifiers perform three functions: they convert AC mains power into DC; they raise the voltage of the input signal; and they drive the speakers from the DC supply with a replica of the increased input signal.  There are separate sections to perform these functions in most amplifiers.

There are two basic approaches to audio amplification: linear and switching.  A linear amplifier contains output devices that provide a smoothly-varying output voltage to drive the speakers.  These devices have to withstand the full difference between the power supply DC voltage and the output signal voltage at all times, so they have to be able to dissipate a lot of heat.

Switching amps have output devices that turn fully on and off very quickly and at high frequency, and it is the average of the on and off times that forms the audio signal for the speakers.  There is a filter between the devices and the speakers.  These output devices dissipate much less heat.  By analogy with the different bias classes in linear amplifiers, these are generically referred to as “Class D.” Note that they are sometimes called “digital” amps, but this is not accurate.  Individual manufacturers have started to use particular class designators to hype their proprietary modulation and filtering schemes.

Switching amplifiers are relatively new and there are not many models available.  They put out little heat compared to linear amplifiers and should be easier to accommodate.  Beyond the need for a good power supply, they also have to deal effectively with the high-frequency electrical noise (in essence, RF) that results from the rapid switching.  If the filter does not fully absorb it, this noise can get into other parts of the audio system and cause problems.

There are many varieties of linear amplifers, as these were historically the first, and remained the only kind used for audio for many years.

Single-Ended Triode (SET)

The first kind of electronic amplifier developed was the single-ended triode.  The triode is a simple kind of vacuum tube with three elements.  Many tube enthusiasts consider it still the best amplifying device for audio.  Because the output tube can not be allowed to turn off in normal operation in this kind of amp (they must be biased in Class A), the tube has to dissipate more power as heat than it can deliver to the speaker.  This means SET amps typically have low output power ratings, and are not suitable for speakers that are less efficient or that have to fill large rooms.  They cannot deliver or absorb high currents.  They have to be exposed so that the heat can be dissipated efficiently.

Tubes contain very thin and light metal parts inside the glass envelopes.  These parts vibrate.  Since the vibration directly affects the spacing of the parts, it modulates the signal.  This property of tubes is called being “microphonic.” All tubes have it to some extent, and the ultimate resolution of a tube circuit depends on measures to reduce the tendency of the tube elements to vibrate.  Transistors have their own problems, and transistor circuits may also be microphonic, so this is not a reason to dismiss all tube amps out of hand.  In fact, the most transparent and resolving amps that I know of use tubes.

SET amps have one other unique feature.  They use a special kind of transformer to connect the output tube to the speaker.  The tube runs at hundreds of volts while the speaker requires lower voltage and higher current, so a transformer is necessary in this kind of amp.  The transformer has to tolerate the DC current through the tube.  This special requirement makes it larger and heavier than other types of audio transformers for the same power level.

The materials and construction methods used to make the output transformer are very important in determining the resolution and bandwidth of the SET amp.  The ultimate resolution, as long as the tube vibration is controlled, is determined by the magnetic grain of the transformer core.  Unfortunately, materials with fine grain are also easily saturated, and give limited bass response.  This means there is a delicate balancing act in designing and building the output transformer, and the amp has to be designed to make best use of it.  The better SET amps are designed by people who understand the output transformer limitations and work closely with the transformer provider.

Push-Pull (tube)

Most tube amps are push-pull with transformer coupling.  In this type of amp, there are two (or two sets of) power output tubes per channel.  As the name suggests, one of them is used to provide most of the current for the positive side of the signal, while the other is used to provide most of the current for the negative side.  The currents through the tubes balance each other through the output transformer, so it does not have to be as cumbersome as for the SET amp.  However, when the signal passes through zero, the magnetic grains in the transformer have to change the direction of their magnetization.  This introduces a subtle distortion and limits the resolution compared to an SET amp, where the transformer core magnetization is always in one direction.

Push-pull tube amps are usually biased in Class AB.  Most listening is done at lower power levels where the amp remains in Class A, so the distortion from tube cutoff is not apparent.  As long as the amp remains in Class A, the power drawn from the supply is constant and independent of the signal because of the opposing actions of the tubes (as one increases its current, the other decreases by the same amount).  This makes the power supply less critical than for SET amps, where the power drawn from the supply varies instantaneously with the signal.  It also allows the output circuit to reject power supply noise and to have more authority than transistor amps with similar output rating.  This is because the transistor amp uses two separate power supplies, and deviations from ideal behavior in each one are more likely to show up as audio distortion.

The output stage of a push-pull amplifier is inherently balanced and differential.  This means amplifiers with single-ended inputs have to convert the signal to differential before the output stage.  The circuit that does this is called the “phase splitter,” as it has to generate a negative copy of the original signal.  Imperfection in this circuit is another limitation on the fidelity of the push-pull amp.

Push-pull tube amp designs can be scaled to deliver more power than SET amps with similar size and heat generation.  The output transformers are smaller, lighter, and cheaper to build.  With use of negative feedback, the current capability can be increased so that a wider range of speakers can be used.  The bias currents through the tubes have to be controlled to keep both sides at the same level.  Unless the amp employs automatic control, this adjustment has to be done by the user from time to time.

Most push-pull amps have several output terminals for different load impedances.  These are derived by tapping the secondary winding.  This means the lower impedance connections do not use all of the secondary winding, and the sound quality may suffer as a result.  It is better to use speakers that match the highest impedance output on the transformer.

The output tubes used in most push-pull amps are called “pentodes” because they have five elements instead of three.  These extra elements were added to tubes to allow them to work at higher (radio) frequencies.  Pentodes have different drive requirements compared to triodes, and the output transformers have to be wound for the particular tubes to be used.  It is generally not possible to swap different tube types, but there are some amps that can accommodate a few similar types without major circuit changes.

Output-Transformerless (OTL)

This is an exotic form of tube amplifier in which the output transformer problems are eliminated.  Many tubes are used in parallel with a relatively low voltage to allow the speaker to be coupled directly to the output stage.  This type of amp has the best ultimate resolution, but cannot deliver much current.  Care and feeding of the tube complement is a challenge and requires a lot of user time and knowledge.  All tubes wear out, but the other kinds of tube amps are easier to maintain.

Transistor, or Solid-State

Transistor amplifiers offer freedom from the headaches of tube fragility, heat, and wearout.  They come in a variety of power ratings.  Because most transistor circuits require lots of negative feedback to reduce distortion, they almost all offer better current capability and speaker damping than comparable tube amps.  Within this class, there are more and less capable amps for high current, so careful research is needed if a true high-current model is required.  The ability to double the power as the load impedance is cut in half indicates a robust power supply but may not guarantee freedom from distortion in the case of speaker-imposed currents. Examine the power rating carefully for the load impedance used.  If the speakers are not the same impedance, or higher, then the amplifier may burn out quickly.  This is because transistors control a lot of current in a small area of silicon, and the leads inside the transistor packages are no bigger than absolutely necessary.  If the load impedance is too low, the transistors might have to deliver more current than they were designed for, and the leads can be melted.  More robust amplifiers will usually state their ability to deliver power into lower load impedances, so the absence of such a statement should be cause for concern.

Transistor amps suffer from a characteristic kind of distortion arising from the nature of the transistors themselves.  Within a transistor, the signal modulates the bandwidth of the circuit stage in which the transistor is employed.  Some people find this kind of distortion to be annoying and the name “transistoritis” has been applied to it.  Only a very few transistor amps have been designed to minimize this problem, and it is an open question as to whether even these can approach a properly-built OTL tube amp for resolution and transparency.

This kind of distortion is common to all types of transistors.  The major types used in audio amplifiers are Bipolar Junction (BJT), Junction Field Effect (JFET), and Metal-Oxide-Semiconductor Field Effect (MOSFET).  Each is designed and manufactured a different way and has different strengths and weaknesses when used in audio circuits.  BJTs deliver high currents but have to be driven by currents and not voltages, and are subject to thermal runaway.  JFETs are linear but small.  MOSFETs can be made large, but have higher resistance than BJTs for a given size.  There are claims that MOSFETs can be made to sound like tubes, but these claims arise from comparison with poorly-designed tube circuits.

It is time to define “capacitance” in order to discuss stability.  Capacitance is a measure of how much electrical charge is stored within an object for a given voltage across it.  It represents stored energy, since the charge carriers repel each other and have to be forced to stay together.  The energy can be recovered when the carriers are allowed to depart.  Circuit components called capacitors have a lot of capacitance for their size, and are used in filters and to block DC.  However, everything has some capacitance associated with it.

Transistor amps are the most likely to suffer stability problems.  Capacitance is a property of all electronic devices.  Each stage in an amplifier has some capacitance at its input.  There will be a time delay, or increase in phase in the signal, caused by the finite resistance of the previous stage or source that supplies the charge (as input current) to this capacitance.  All these time delays through the amplifier add up.  Since most transistor amps use negative feedback, there is a stability problem if the signal at the output is delayed long enough to turn it into positive feedback when it gets to the input.  Speakers such as electrostats or exotic cables with a lot of capacitance can make a stable amp become unstable.  Before using an amp with electrostats or with specialized high-capacitance cables, it is wise to discuss the stability issue with the amplifier manufacturer.  Some will void the warranty if their product is used under these conditions.

A tube amp could be designed to be marginally stable.  Most tube amp designs have been worked out over many years, and they use less feedback, so it is less likely for a user to encounter one.

Transistor amps usually exhibit hard clipping when driven beyond their rated capacity.  This can be dangerous for tweeters, so it is important to use an amp of sufficient power output when driving low-efficiency speakers.

Power supply electrical noise can be a major problem with powerful transistor amps.  This is why it is not always advisable to use the biggest amp money can buy.

Hybrid

There are some power amps with tube input stages and transistor output stages.  The intent is to avoid the heat, bias adjustment, and transformer problems of tube output stages while preserving some of the clarity of the tube voltage amplifier stages.  This kind of design also reduces the amount of feedback required, so it may be more stable.  Be careful if there is any indication that the designer considers tubes to be a source of euphonic coloration: this means the design is inadequate.