Sketches: Normand Daigle
We have finally arrived at the last installment of this series on power amplifier classes. After covering the most common ones — Classes A, B, and AB — we will now turn our attention to the others. Some will only be briefly mentioned, such as Classes C, E, and F, as they are not suited for audio reproduction. Others, like Classes G and H, will be discussed even though they have seen limited use. Finally, we will take a closer look at Class D, a technology that has been around for quite some time but has only recently gained recognition and is currently revolutionizing the Hi-Fi world. As with Classes B and AB, the primary goal of developing these other amplifier classes was to improve efficiency, but as we will see, this comes at the cost of increased design and manufacturing complexity.
As previously mentioned, the power dissipation at the output transistors of a power amplifier is the result of the supply voltage multiplied by the current flowing through it. In Class A, the static current is very high, leading to a very low efficiency of around 30 %. In Class AB, the current is just enough to eliminate crossover distortion — just a few dozen milliamps — so dissipation is much lower, making losses tolerable for most low- to moderate-power applications. However, when high power is required, the supply voltage must be proportionally increased, inevitably increasing power dissipation. This issue is further compounded by the need to use multiple pairs of transistors in parallel to handle the resulting high currents, further increasing dissipation. Additionally, designing high-power amplifiers requires compromises that can negatively impact sound quality.
The nature of musical content itself offers an opportunity to improve efficiency. Music reproduction is not a static process but a dynamic one, meaning the signal intensity constantly varies. A large portion of the signal is relatively low in intensity, with occasional much larger peaks. The ratio between these can reach 20:1 in uncompressed recordings. One elegant way to minimize power dissipation in high-power amplifiers while maintaining sound quality is to handle these two power bands separately using a switching power supply system at the output (not to be confused with the switching power supplies we discussed earlier). In such a configuration, the output stage consists of a low-power Class AB section operating on a low-voltage power supply and a high-power section running on a much higher voltage. This is the basic principle of Classes G and H.
The following diagram illustrates the basic output stage of a Class G amplifier. To simplify the presentation, the input section and negative feedback are not shown, but they would resemble what we have seen in other amplifier classes. Notice that the respective collectors of transistors Q2 and Q3 are powered at + 20 V and – 20 V via isolation diodes D2 and D3. These two transistors form the amplifier’s low-power section, capable of delivering about 20 W RMS into an 8 Ω speaker. Meanwhile, the collectors of Q1 and Q4 receive a supply of + 40 V and – 40 V. Since power is calculated as the square of the signal voltage divided by the load impedance, doubling the supply voltage allows for a fourfold increase in maximum available power — approximately 80 W RMS. With a low-intensity signal, the amplifier operates exclusively with Q2 and Q3 using the ± 20 V supply. However, if the output signal exceeds the level determined by Zener diodes Z1 and Z2, transistors Q1 and Q4 activate, enabling maximum power output. Under normal conditions, the idle current (bias) only flows through Q2 and Q3, meaning idle dissipation is determined by the ± 20 V supply. The circuit shown is, of course, highly simplified, and transitioning between power supplies poses complex design challenges to avoid introducing distortion during switching.
Class G Principle
As seen in the following diagram, Class H is essentially a variation of Class G, where the power supply change is continuous and progressive rather than a simple switch. In this case, once the low-power section’s level is exceeded, the high-power supply voltage is modulated by the signal. The result is that the power supply always provides just enough voltage to reproduce the audio signal while keeping dissipation in Q1 and Q4 to a minimum. The circuit closely resembles that of Class G, with the key difference being that transistors Q1 and Q4 are now controlled by a signal proportional to the intensity of the audio signal. The main challenge in designing such a circuit is ensuring that the power supply modulation precisely follows the audio signal.
Class H Principle
Although Class G, and even more so Class H, improves efficiency in high-power amplifiers, this comes at the cost of increased complexity. Furthermore, the reduction in power consumption is not particularly significant. The power supply is more elaborate, the circuit contains more components, and the design is more challenging. Despite this, some manufacturers continue to pursue this technology.
And finally, the elephant in the room! There is no doubt that Class D is currently the most discussed topic in the amplifier world — and for good reason. A well-designed Class D amplifier can rival the best linear amplifiers on the market. With over 90 % efficiency, affordable prices, and incredibly low weight and size, they are an ideal choice for serious audiophiles. However, not all Class D amplifiers are created equal, and determining which ones truly perform well can be challenging. Let’s start by understanding how this technology works.
Unlike the other amplifier classes we have analyzed, Class D amplifiers do not operate in linear mode but instead use switching technology. The principle is similar to the switching power supplies that have been common in computing for years and are increasingly used in consumer electronics. Essentially, instead of controlling the power stage in a linear fashion, variable-width pulses are used — hence the term Pulse Width Modulation (PWM). It is important to note that Class D is not a form of digital amplification. It does not involve decoding a digital audio signal where each word represents a sample of the original analog signal taken at a specific interval (e.g., 16-bi t/ 44.1 kHz).
The following functional diagram illustrates the different components of a Class D amplifier. The modulator receives the analog audio signal and a high-frequency triangular wave from an oscillator — typically 15 to 20 times the highest frequency to be reproduced, meaning between 300 and 400 kHz, and sometimes beyond 1 MHz. If we take the latter as an example, the modulator generates a pulse every microsecond with a width proportional to the waveform’s intensity at that moment. The result is a train of pulses with varying durations and low amplitude, which are then sent to the switching controller. This controller uses high-speed power MOSFETs to switch the power supply and produce an output train of pulses similar to the originals but with an amplitude matching the supply voltage. The resulting signal is then routed to the speaker via an LC filter consisting of a ferrite-core inductor placed in series, followed by a capacitor in parallel with the load. The effect of the filter is to eliminate all the steps that link one pulse to the next and produce a perfect reconstruction of the original analog wave.
Principle of Class D
In the previous diagram, we can see that the power supply for the switching controller is symmetrical, meaning it has both a B+ and a B-, which allows for direct coupling with the load. However, just like linear amplifiers, a Class D amplifier can be built with a single power supply where B- is essentially ground, but this implies that there will be a DC (direct current) component equivalent to ½ B+ at the output. In such cases, a BTL (Bridge-Tied Load) configuration will be used, meaning two amplifiers reproducing the same signal will each be connected to a terminal of the load, as shown in the following image. By inverting the phase by 180° on one of the amplifiers, as indicated by the dot at the output of the lower module, the output voltage is doubled, and consequently, the potential power is quadrupled. In this configuration, the speaker will no longer be connected to ground but will have ½ B+ at each terminal, resulting in 0V DC.
It is also possible to use a PBTL (Parallel Bridge-Tied Load) configuration, where modules are added in parallel with others to increase current capacity, allowing for lower impedance usage and increased power. This is why some Class D amplifiers are advertised as capable of handling a 2 Ω load. In PBTL mode, it is possible to achieve nearly 40 W on 2 Ω with a simple 12 V DC power supply. On this point, it is important to note that the power ratings advertised by manufacturers of budget Class D amplifiers are often highly exaggerated. In most cases, they refer to peak power, combining both channels at 2 Ω. At best, the actual power per channel at 8 Ω will be roughly a quarter of the stated value.
BTL Configuration
One of the key factors in the performance of a Class D amplifier is its power supply, and the requirements are very different from those we are used to with linear amplifiers. Without delving too deeply into this topic, it is important to mention that the output stage of a Class D amplifier tends to return current to the power supply, creating a bus pumping effect, which can unpredictably increase the supply voltage. Therefore, this characteristic must be considered when designing the power supply, whether it is single or symmetrical, linear or switching. It is worth noting that using the BTL or PBTL configuration helps to minimize bus pumping by sharing the workload between two amplifiers operating in opposite phases.
The budget versions of Class D amplifiers use integrated circuit (IC) chips that contain the output stage. These are mainly found in portable and entry-level devices. Many of these lack feedback control, which significantly limits their performance. More advanced models use external MOSFETs, allowing for much higher currents and supply voltages, and they incorporate global feedback for significantly better power and performance.
Given everything we’ve discussed, it would seem that Class D amplifiers are the ideal solution for audio reproduction, and they are indeed becoming the dominant technology. However, they do have their drawbacks, such as the complexity of PCB (Printed Circuit Board) design, which is highly critical, and the RF (radio frequency) interference they can generate due to their high switching frequency. Nevertheless, there is no doubt that they are the way of the future. If designed with care — something that is not within the reach of every designer and explains why most manufacturers use OEM (Original Equipment Manufacturer) modules — their performance is exceptional. Their excellent efficiency, low weight, compact size, and continuously decreasing manufacturing cost make them an essential solution.
Looking ahead, the next step will be the increased availability of Class D amplifiers capable of directly receiving digital signals (S / PDIF) from sources without first converting them to analog. Such devices currently exist but are not yet widespread.
This brings us to the end of this series of articles. Hoping that it has both
informed and entertained you, I wish you great HiFi adventures!
For referehnces… Part 4 Part 3 Partr 2 Part 1
