Amplifier Classes

Alan March

Founder of March Audio
Staff member
As I'm sure you know March Audio amplifiers are Class D designs, but what does this mean? Its commonly assumed that the "D" stands for digital, but this is not the case, they are actually very analogue 😀.

So I thought it would be helpful to post some information on the various classes of amplifier, how they work and why we think class d is the way forward.

I previously posted the following articles elsewhere and a lot of the content has been plagiarised but there is no point in re-inventing the wheel 😁
 
Amplifier Classes

1648979214639.png


Not all amplifiers are the same and there is a clear distinction made between the way their output stages are configured and operate. The main operating characteristics of an ideal amplifier are linearity, signal gain, efficiency and power output but in real world amplifiers there is always a trade off between these different characteristics. Amplifier Classes is the term used to differentiate between the different amplifier types.

Amplifier classes are mainly lumped into two basic groups. The first are the classically controlled conduction angle amplifiers forming the more common amplifier classes of A, B, AB and C, which are defined by the length of their conduction state over some portion of the output waveform, such that the output stage transistor operation lies somewhere between being “fully-ON” and “fully-OFF”. You can see in the diagram above how much a transistor conducts of a sine waveform for each class.

The second set of amplifiers are the newer so-called “switching” amplifier classes of D, E, F, G, S, T etc, which use pulse width modulation (PWM) to constantly switch the signal between “fully-ON” and “fully-OFF” driving the output hard into the transistors saturation and cut-off regions.

Transistors

Its difficult to know how much detail to go into, but I think its probably beneficial to first briefly talk about transistors.

Transistors are fundamentally three-terminal devices. On a bi-polar junction transistor (BJT), those pins are labeled collector (C), base (B), and emitter (E). The circuit symbols for both the NPN and PNP BJT are below:

1648979241345.png

Transistor Construction

Transistors rely on semiconductors to work their magic. A semiconductor is a material that's not quite a pure conductor (like copper wire) but also not an insulator (like air). The conductivity of a semiconductor -- how easily it allows electrons to flow -- depends on variables like temperature or the presence of more or less electrons. Let's look briefly under the hood of a transistor. Don't worry, we won't dig too deeply into quantum physics.

A Transistor as Two Diodes

Transistors are kind of like an extension of another semiconductor component: diodes. In a way transistors are just two diodes with their cathodes (or anodes) tied together:

1648979282024.png

The diode connecting base to emitter is the important one here; it matches the direction of the arrow on the schematic symbol, and shows you which way current is intended to flow through the transistor.

The diode representation is a good place to start, but it's far from accurate. Don't base your understanding of a transistor's operation on that model (and definitely don't try to replicate it on a breadboard, it won't work). There's a whole lot of weird quantum physics level stuff controlling the interactions between the three terminals.

The transistor is kind of like an electron valve. The base pin is like a handle you might adjust to allow more or less electrons to flow from emitter to collector. Let's investigate this analogy further..

The Water Analogy

We can say that current is analogous to the flow rate of water, voltage is the pressure pushing that water through a pipe, and resistance is the width of the pipe.

1648979307944.png

Unsurprisingly, the water analogy can be extended to transistors as well: a transistor is like a water valve -- a mechanism we can use to control the flow rate.

There are three states we can use a valve in, each of which has a different effect on the flow rate in a system.

1) On -- Short Circuit
A valve can be completely opened, allowing water to flow freely -- passing through as if the valve wasn't even present.

1648979335553.png

Likewise, under the right circumstances, a transistor can look like a short circuit between the collector and emitter pins. Current is free to flow through the collector, and out the emitter.


2) Off -- Open Circuit
When it's closed, a valve can completely stop the flow of water.

1648979355298.png

3) Linear Flow Control
With some precise tuning, a valve can be adjusted to finely control the flow rate to some point between fully open and closed.

1648979377877.png

A transistor can do the same thing -- linearly controlling the current through a circuit at some point between fully off (an open circuit) and fully on (a short circuit).

From our water analogy, the width of a pipe is similar to the resistance in a circuit. If a valve can finely adjust the width of a pipe, then a transistor can finely adjust the resistance between collector and emitter. So, in a way, a transistor is like a variable, adjustable resistor.

A transistor controls a larger current (from the power supply) with a much smaller current (from the audio signal). A small current flowing in the base controls a larger current flowing through the collector emitter. This is how it is used to amplify.

So now you are a transistor expert lets move on to the amplifier classes.

Next post will be about Class A
 

Attachments

  • 1648979191591.png
    1648979191591.png
    9.3 KB · Views: 0
Class A Amplifier

Firstly we need to understand the concept of bias. Transistor Biasing is the process of setting a transistors DC operating voltage or current conditions to the correct level so that any AC input signal can be amplified correctly by the transistor. Sort of how much we turn the tap on and how much water flows when we have no input signal. The input audio signal will then turn the tap on more, or close it further.

Class A Amplifiers are biased around the Q-point within the middle of its load line and so is never driven into its cut-off or saturation regions thus allowing it to conduct current over the full 360 degrees of the audio input cycle. As the DC voltage is sat in the middle of the range the signal can go lower and higher. So In this example the one transistor amplifies the whole waveform. The output transistor of a class-A topology never turns “OFF” which is one of its main disadvantages.

1648979429059.png

So as any good Audiophile knows this topology can be made very linear (although its not a given), however it has a massive disadvantage. It's extremely inefficient, maybe only 25%. The transistor is always conducting, even when there is no audio signal, the tap is always open flowing water. As a consequence it gets hot and wastes lots of electricity. Also due to this inefficiency it is more challenging and expensive to create high power amplifiers.

Class B next
 
Class B amplifiers were invented as a solution to the efficiency and heating problems associated with the previous class A amplifier. The basic class B amplifier uses two complimentary transistors either bipolar of FET for each half of the waveform with its output stage configured in a “push-pull” type arrangement, so that each transistor device amplifies only half of the output waveform.

In the class B amplifier, there is no DC base bias current as its quiescent current is zero, so that the dc power is small and therefore its efficiency is much higher than that of the class A amplifier. However, the price paid for the improvement in the efficiency is in the linearity of the switching device.

1648979460550.png

When the input signal goes positive, the positive biased transistor conducts while the negative transistor is switched “OFF”. Likewise, when the input signal goes negative, the positive transistor switches “OFF” while the negative biased transistor turns “ON” and conducts the negative portion of the signal. Thus the transistor conducts only half of the time, either on positive or negative half cycle of the input signal.

Then we can see that each transistor device of the class B amplifier only conducts through one half or 180 degrees of the output waveform in strict time alternation, but as the output stage has devices for both halves of the signal waveform the two halves are combined together to produce the full linear output waveform.

This push-pull design of amplifier is obviously more efficient than Class A, at about 50%, but the problem with the class B amplifier design is that it creates distortion at the zero-crossing point of the waveform due to the transistors dead band of input base voltages from -0.7V to +0.7.

I dont thnk you ever see a class B audio amp due to this.

Crossover distortion

1648979486265.png
 
Class AB Amplifier

As its name suggests, the Class AB Amplifier is a combination of the “Class A” and the “Class B” type amplifiers we have looked at above. The AB classification of amplifier is currently one of the most common used types of audio power amplifier design. The class AB amplifier is a variation of a class B amplifier as described above, except that both devices are allowed to conduct at the same time around the waveforms crossover point reducing the crossover distortion problems of the previous class B amplifier.

The two transistors have a very small bias voltage, typically at 5 to 10% of the quiescent current to bias the transistors just above its cut-off point. Then the conducting device, either bipolar of FET, will be “ON” for more than one half cycle, but much less than one full cycle of the input signal. Therefore, in a class AB amplifier design each of the push-pull transistors is conducting for slightly more than the half cycle of conduction in class B, but much less than the full cycle of conduction of class A.

In other words, the conduction angle of a class AB amplifier is somewhere between 180o and 360o depending upon the chosen bias point as shown. One transistor deals with the positive half of the cycle and the the other transistor with the negative half of the cycle. The small amount of overlap helps reduce the crossover distortion.

Class AB Amplifier

1648979525003.png


The advantage of this small bias voltage, provided by series diodes or resistors, is that the crossover distortion created by the class B amplifier characteristics is overcome, without the inefficiencies of the class A amplifier design. So the class AB amplifier is a good compromise between class A and class B in terms of efficiency and linearity, with conversion efficiencies reaching about 50% to 60%.
 
Class D Amplifier

In Classes A, B and AB, the problem is lack of efficiency. Some power is wasted, and we would prefer that it could be sensibly employed in driving the loudspeakers to ever-higher sound pressure levels — or, at least, not converted to heat. Where power is wasted is where a transistor is in partial conduction. When a transistor is fully conducting, it's like a piece of wire, and a piece of wire loses hardly any power. When a transistor is fully off, it doesn't conduct at all, and if it doesn't conduct at all, there's no power to waste. It's the in-between stages that cause the problem, where the transistor wastes power and gets hot. So what if we could find a way for transistors to be used only in their fully-on or fully-off states. If that were possible, no power would be lost. But is it possible...?

1648979587718.png

It is, and the solution is what we call Class-D. The diagram above shows a simplified Class-D amplifier. First, let's look at the similarities between this and what we've already discussed. You can see two transistors, in push-pull configuration, as before. The transistors look slightly different because they are MOSFETs (Metal Oxide Semiconductor Field Effect Transistors) rather than 'ordinary' transistors. The output transistors have to be fast, so that they can switch very quickly between fully on and fully off. It also helps for the on and off states to be 'really' on and off. The closer the transistors can get to full conduction or full non-conduction, the greater the efficiency of the amplifier will be. Clearly, though, as well as the similarities there are some differences.

Let's start at the output. What's not going to happen is that the transistors create a high-voltage version of the input signal. What is going to happen is that they switch alternately to lift the output all the way up to the positive supply rail, then all the way down to the negative supply rail, as quickly as possible, with no in-between voltages. This is clearly going to be a pulse waveform. Now here's the clever bit: if the width of the pulses can be made proportional to the input signal's instantaneous level, the power delivered to the loudspeaker, averaged over time, will be the same as if the input signal had been amplified in the conventional way. Think about that for a moment, because it's key to how a Class-D amplifier works.

Going further, if the output is filtered to remove the high frequencies and sharp corners of the pulse waveform, the original input signal will be reconstructed, exactly the same shape as it was, but bigger. The net result is an amplified signal that you couldn't distinguish from that produced by a conventional Class-AB power amplifier.

1648979609390.png

But how is the pulse waveform produced? OK, it isn't simple, but it isn't rocket science either. First we need a circuit building-block known as a comparator. A comparator has two inputs: let's call them Input A and Input B. When Input A is higher in voltage than Input B, the output of the comparator will go to its maximum positive voltage. When Input A is lower in voltage than Input B, the output of the comparator will go to its maximum negative voltage. Figure 7 shows how the comparator operates in a Class-D amplifier. One input (Input A in my example) is supplied with the signal to be amplified. The other input (Input B) is supplied with a precisely generated triangle wave. When the signal is instantaneously higher in level than the triangle wave, the output goes positive. When the signal is instantaneously lower in level than the triangle wave, the output goes negative. The result is a chain of pulses where the pulse width is proportional to the instantaneous signal level. Magically simple! We call it 'pulse width modulation', or PWM. And that's all there is to it. You now understand how a Class-D amplifier works, and if anyone tries to pull the wool over your eyes and convince you that the 'D' stands for 'digital', you can tell them how wrong they are, with confidence. Class-D is not digital.

So it sounds simple in theory, but in practice it does have challenges that need to be overcome which may in the past have led to a poor reputation for class D. However these challenges have now been overcome. There are also a number of myths that need addressing. I will talk about these in another post.
 
Class G Amplifier Probably the last Amplifier class worth mentioning is class G. It is essential a class A/B amplifier with one difference. To increase efficiency and reduce heat it has multiple power supply voltages. When the input signal is small it runs on the lowest supply voltage. As the input signal gets larger and it is required to output a larger signal it seemlessly switches to the next higher supply voltage and so on. This means that the minimum of voltage is dropped across the output transistors, minimising power losses land heat. Seems like a good idea but care has to be taken to ensure the changes of supply voltage are fast enough to follow musical transients, and don't create noise.
 
Back
Top