With the risk of making enemies right away, I am committed to clarifying a few things.
We are used to classifying audio amplifiers according to their working class.
This classification in itself is not wrong and is surely something to consider when we start evaluating audio amps. However, there are certain facts that are worth looking into with a bit more accuracy:
- amplifiers DO NOT have a working class, only the final stage does
- the claimed class isn’t always right, sometimes it could even not make sense at all
- a class doesn’t guarantee a good performance, it depends on how the circuit is designed and on which components are used
Since using electronics to explain electronics is not a good method – those who don’t have a good background won’t be able to follow – it’ll be better to use an analogy that almost everyone will understand.
Claiming that an amplifier has a certain working class is sort of like affirming that our car runs on gasoline, diesel or is electric. We can’t blindly say that an engine that runs on diesel is better than one that runs on gasoline, because there are well built gasoline engines that perform better than badly built diesel engines, and vice versa.
Although it is true that the engine type, like the working class of a final stage of an amplifier, has some distinctive traits. For example, an excellent gasoline engine converts about 25% of the fuel’s chemical energy in mechanical energy – this percentage is called efficiency – while an excellent diesel engine can go up to 40%.
Notice the adjective “excellent”, because we’re talking about a state-of-the-art realization.
Making things worse is always easy, there are plenty of examples. Instead, there are limits in making them better.
So we can’t say that there is a “better” engine for ordinary use, because if you want the best performance, as in Formula 1, you have to choose a gasoline engine – because with the same engine size you have more horse power – instead if you want minor consumption with the same horse power, a diesel engine will be the right choice. Other things to take into consideration when you’re not a F1 pilot are maintenance costs, taxes, pollution and so on, which will influence the choice to make.
Similar observations are also true for audio amplifiers. For example when we consider efficiency: a class A has a maximum efficiency of 25%, which means that in order to drive the loudspeakers with 25W the final stage will have a power drain of 100W and the remaining 75W will be dissipated as heat – this explains the size or the generous cooling fins – while a class AB has an efficiency between 50% and 75% and a class D can go up to 90%.
Let’s look at things “from above”: an amplifier takes in ENERGY and INFORMATION and lets out ENERGY DRIVEN by INFORMATION.
An amplifier, conceptually, is an object with two inputs and one output.
We need this “energy modulation” because information alone isn’t able to influence physical phenomena, as you can easily verify by taking a picture of your car and trying to get on it to go somewhere: you’ll be disappointed.
Sole information isn’t enough because we live in a physical world and, to move physical objects, we need energy. This is the reason why we don’t connect the output of a CD to loudspeakers: the speakers wouldn’t move and therefore would not move the air.
On the other hand, energy alone isn’t useful if not “tamed” by information, like gasoline in a tank if we set it on fire instead of burning it correctly inside the engine: we won’t get useful motion but a dangerous fire. A car will fulfill its purpose if it has fuel, energy, and information, a thinking human being capable of driving it.
Therefore, a car too is an amplifier, like many other machines, and for example our muscles are one as well. Nature has created amplifiers long before we have, and moreover without arguing about classes.
Efficiency is the ratio between the amount of energy that comes out and the amount that goes in, and if 100 goes in and 60 comes out, the difference is almost always lost as heat – or as various forms of radiations – and this is never a good thing, because that 40 of “untamed” energy goes around making trouble, as all energy out of control does.
As the efficiency increases, under equal consumption and dimensions we can attain more power, like in engine displacement, which doesn’t mean that 300W obtained with a class D cannot be obtained with the other classes, but to reach them we will have to increase the volume, so if we obtain our 150W+150W with a high efficiency class we will have a small device that will consume a little over 300W, whereas if we want to obtain them with a low efficiency class, as is class A, we’ll find ourselves having a big, heavy, expensive final stage that consumes a lot even with the volume turned down to minimum.
Amplifiers of equal power, 2x150W, compared in terms of class, dimensions and weight. Device choice is entirely illustrative.
But an audio enthusiast isn’t as passionate about power consumption as he/she is about fidelity and, if to obtain with our loudspeakers the desired sound we have to convince our wife or husband that the expensive and voluminous amplifier we placed in our living room was the only possible choice, we’ll take the risk.
Then let’s see the details: what’s the difference between the different classes in terms of quality?
Here things are more complicated than a simple percentage.
There are a lot of different ways to make an amplifier and, as with engines, the choice of architecture and components changes things and may also turn them around. Especially if one follows ideologies and the trend of the moment instead of technology.
The letters that identify each class are not a “grade” like the ones they give in school, where A is the maximum, B is a good but lower mark, C is just enough and so on. The reason why these classes are called as they are is substantially historical: the letters have been assigned based on the order of appearance of the various architectures.
Class A final stage
The first to come out was class A, which didn’t even have a name since it was the only one, and was built with triods, because only those were available. We’re talking over 100 years ago, at the beginning of the Twentieth Century.
Audion, the first triode. Valves are vacuum tubes.
These pioneering amplifiers were made of a single vacuum tube that amplified the entire signal, with a very low efficiency. As for the output transformers that we always see on audio valve amplifiers, they represent a problematic factor and are often unnecessary. They become indispensable if we only have one tube and have to drive loudspeakers, though there are devices that do well without them – the so called OTL, Output Transformer Less – while in other implementations that are not audio related, they are seldom used.
Class A final stage
This first final stage class works on the signal “as a whole” and its distortion, which is simply the difference between the signal that enters and the one that comes out “energized”, in theory is really low but depends entirely on the quality of the component. A beginner thinks of components as ideal, perfect, but they actually never are. So having a class A does not guarantee fidelity.
Class B final stage
In a class B final stage, which is so called because B is the letter that comes after A, only half the signal is amplified, the upper 50% or the lower 50%, with a higher efficiency but a greater distortion, since what comes out has a very different shape compared to the entering signal as only half of it is reproduced.
Class B final stage
With the arrival of class B, a name was also given to the class that came before it, class A, to distinguish them. Sometimes saving energy is more important than distortion, and the fact that with the same consumption the obtainable power could even be three times that of an A class amplifier made this working mode very desirable in fields that were not High Fidelity.
Class AB final stage
Soon it was noticed that since a class B final stage amplified only half the signal, one could use another identical final stage to amplify the other half, and so the AB class was born. In the circuit technique the combination of two complementary power amps is more complex than an educational treatment can show, but the main concept is that one stage has the task of handling the negative part of the signal and the other has to handle the positive part. The resulting system is composed of two properly modified B sections.
AB class final stage, simplified idea, in reality things are more complicated
The consequences of this combination are:
- for classes A and B only one final stage is necessary while for class AB you need two, and this complicates things because it means double the components, double the costs and double the size, especially when it comes to vacuum tubes
- there’s a region where one final stage stops working and the other starts amplifying, called “crossover region”, and this complicates things as well because if everything is not calibrated to perfection, we could have distortion and thus loose signal quality
Class AB has a better efficiency and this allows us to have good power amplifiers that are not big and heavy, but in order to have a good performance the two amplifiers (one for each half of the signal) have to behave exactly in the same way and have to be stable while working. This is hard to obtain using valves, which are inherently unstable due to temperature variations, aging, production and storage methods and, sometimes, even... the owner’s mood.
If the two parts of the signal are not treated in the exact same way and if there’s not an accurate control of the crossover zone, what comes out has different shape than what goes in and this is distortion. The management of how the finals behave is called "polarization", but its description is left to more technical articles.
The image above shows the typical crossover distortion in a class AB final stage when the polarization is not correct.
The easiest and most reliable thing to do using vacuum tubes is to work in class A, and that’s the reason why most valve amps are in this class.
When components offering good accuracy and stability have started to be available, distortion proper to class AB has been minimized with very good results, which is why almost all AB final stages are solid state.
Class C final stage
Class C, so called because it came after B, is a class where less than half the signal is amplified, typically 35-45%, but with an even higher efficiency which comes at a price, the distortion is in fact very high. Think how it would look like if we had a portrait where only a third of the face has been drawn and the rest is… at random... That’s why it is used only in single frequency applications, which definitely do not concern audio reproduction.
Class C final stage
If we put two C stages together we’d get something useless, because part of the signal would never be amplified, as 40+40 doesn’t make 100, which is why you’ve never heard of class C in audio.
Class D amplifiers
Class D, so called because it came after C, is a type of “impulse” amplification and no, it is not digital, although the letter can be misleading. Don’t even think about it.
Pulse amplifiers are analog, like all other amplifiers, because digital ones don't exist yet and will unlikely exist unless the meaning of the word "digital" is changed. Until then, "digital amplifier" will only be a phrase with the advertising purpose of evoking something modern, not even that much anymore, since digital electronics has existed for over 70 years.
Class D amplifier. Unlike the previous ones, class D is an architecture and not a type of final stage.
Digital clarification
Let’s spend a few more words on the definition of digital because there’s a lot of confusion, surprisingly even among insiders.
For something to be called "digital" it must be in numerical form, i.e. translated into information represented by numbers.
Class D amplifiers don't process numbers, they produce pulses, and their outputs aren’t "stepped" as some mistakenly claim, because they can assume all possible values, like other analog signals. There are no "jumps" as some say, evidently without ever having measured the output of a pulse amplifier.
The fact that a device works in an "ON-OFF" mode does not make it digital, otherwise the Bakelite switch that my grandmother used to press to turn on the first incandescent bulbs a century ago would have been digital. And I can assure you it wasn't.
Digital electronics "uses" on/off states to translate numbers that will then be processed by special circuits through programs, but a power amp that closes and opens is not a digital circuit, so not even a class D amplifier is digital, in fact it can also be built using valves, and even without output transformers.
One could do digital electronics in three, four, nine or eleven logical states, not necessarily two – zero and one – which to many seem offensive because they identify them with something categorical, and so act categorically themselves.
To be honest, until the 1950s computers were made with vacuum tubes as well, so the valve = analog combination is completely false. A component “is” not digital, at most it "can be used" in a digital environment, but only if it’s working on the processing of information in numerical form.
Digital = numerical, let’s never forget it, it’s easy. The rest is analog. Or magic, according to certain advertisements...
Classes continue
Furthermore, there are classes that are not real working classes and this is a widespread problem we humans have with names: many things are called in ways that don't match anything – think about politics – even if here, at least, it hasn't been done with dishonesty.
Class G amplifier, source Wikipedia
So, for example, there is a class G which is not a real final stage class but a way to supply power to an output stage operating with a real working class.
A quick explanation: to produce larger signals, if the impedance is more or less constant – let's think of the 8ohm of our speakers – you need higher power supply voltages to the final stages. On an 8ohm loudspeaker, to deliver power at 5W you need a power supply of at least 20V – class A, AB without output transformer – while to deliver power at 100W you need more than 80V.
In amplifiers that are powered with low voltages, such as those for cars, there are special circuits that raise the battery voltage to those needed in the final stage.
In household or professional amplifiers, it is the mains power supply circuit that provides the right voltages and, as we can imagine, it is as essential a part of the amplifier as fuel for an engine or food for a living being.
The energy dissipated in heat by the amps increases a lot with the power supply voltage – classes A, B, AB – and this means that if you are listening to 5W music with a 100W amp, the heat dissipated and therefore the energy waste is much higher than if you were listening to these 5W with a 20W amp of the same class, because the former is served by a voltage unnecessarily high, if you don't turn up the volume.
A bit like going slow with a super sports car with a huge engine: it will consume much more than a small car at the same speed and weight.
Well, class G is just powering a final stage with different voltages depending on the power that’s being requested. Generally, two or three voltages are employed, lower when the volume is low, higher when the volume is high, in order to have the right voltage at the right time. A power amp "with gears", in short: you start in first gear, then if you turn up the volume above a certain threshold you put the second one, and so on. All to reduce consumption and heat dissipation.
So class G only makes sense if it is combined with a power stage that works according to a true class and then we have a GA or a GAB or a GD, although the efficiency of pulse amplifiers is such that it does not justify these power supply solutions.
Class H amplifier, source Wikipedia
Class H does the same thing as G but in a continuous way, that is it doesn't have two-three-four levels of voltage to be exchanged as needed, but it varies in a continuous and progressive way to provide the best supply voltage to the power amps needed at the moment.
As with G, class H makes no sense on its own, so it must necessarily refer to an actual amplification class or talking about it has no meaning.
Saying "I bought a class H" is like saying "I bought a car with an automatic transmission", without specifying whether it runs on methane, gasoline, diesel, kerosene or potato alcohol. Usually the devices with a class H label are AB units on which they want to save on the size of the cooling fins, and the fidelity is always that of the intrinsic working class, sometimes even worse, because the continuous variation of the supply voltage on an A or AB power amp can cause an increase in distortion.
Be careful: labels don’t make a device; its real characteristics do.
We also find misleading labels that deceive us by mixing the working classes of the power amp with those of other circuits: for example, preamplifiers. This is not correct from a technical – and ethical – point of view because the nameplate class is exclusively that of the final stage, the one that amplifies signal power before the output.
A 1956 Fiat 500 is not an "electric car" because there is an electric starter motor and spark plugs, in fact we correctly call it a "gasoline car". Gasoline combustion is the way in which chemical energy is transformed into mechanical energy. To be called an "electric car" the engine which makes the vehicle move should be completely electric, but it is not.
Similarly for amplifiers: every circuit that is not concerned with handling power, including those found in sources such as CD players, tape recorders, DVDs, digital sources and in general everything that is not exclusively mechanical, is almost always in class A, because it is easier to make and control, and it doesn’t matter if the efficiency is low because the power involved is negligible.
When you adjust the volume, there are small class A integrated circuits – called operational amplifiers – that bring the signal to the right level for the next circuit, and so on up to the final stage.
The tone controls are in class A. The active filters that separate frequencies before amplification are in class A. If you have a preamplifier – valve or solid state – this is almost certainly class A.
According to some people's "view", then, if an integrated amp has a separator – the so-called buffer – or a class A input preamp and a class D output stage, then it would be a "class AD".
Unfortunately, this class does not exist, because the final stage is only D.
Extraudio amplifiers in the declared class… AD
Reasoning this way, since all non-final stages of all amplifiers are class A, we should read on the labels "AAB" instead of "AB", or, if we have a class A power amp with a class A preamp and a source with class A circuits inside, "AAA". I wouldn't want an amp pretending to be an ad.
It is a notation that at best is useless, but above all it is technically incorrect and also very confusing.
It is not good to identify a device – which can be made in many ways – with only the working class of a part of its final stage.
Personally, I have no idea how many types of flour exist, but it’s certainly not with just the small number on the packaging that I could become an expert in cooking.
Understandably, when we do not know the arguments well, it is human to rely on categories that are easy to identify to avoid getting lost among technicalities and data that we would not understand.
We all do it, at the bar, when we talk about politics, or society, or sports we've never played. Simplifications are sometimes useful, but they often generate chaos and nonsense. Creating chaos is not a good result, for a simplification.
Every amplifier as a whole is a "class A" amplifier
If we look at it from the outside – a sort of "black box" that takes a signal and a power supply and thus generates a power output – a High Fidelity amplifier is always... class A, because the output signal is a 100% amplification of the input signal. Some manufacturers don't even write the working class anymore, go check.
What really matters is whether the output signal is the same shape as the input signal, or not, and this depends on many factors, including your setup...
End of first part – To the next article in this series
