Class D, the misnamed digital amplification

The popularization of so-called "digital amplification", a misnomer, has brought about a drastic change in the perception of what amplifiers and, by extension, A/V receivers are. If until recently it was possible to judge an amplifier by its weight, we are now no longer surprised to come across high-power systems that weigh barely 10 kilos.

This Class D amplification has an edge over traditional Class AB designs (the most widely used until now) thanks in part to its high efficiency, which reduces energy losses, so these amplifiers need fewer heat sinks or fans, while at the same time reducing the overall weight of an amplification device. In addition, their small size and lower power consumption simplify the design of any amplifier, saving space and lowering costs. Technically, they also achieve a better signal-to-noise ratio (around 100 dB) and really low total harmonic distortion (THD). Even their levels of electromagnetic compatibility make them suitable for meeting the forthcoming European requirements on electromagnetic pollution.

But what is Class D? Why do we speak of digital amplification? What do switch-mode power supplies have to do with Class D? What drawbacks do they have compared with the traditional Classes A and AB? Why are receivers and amplifiers with different classes of amplification still being made? We answer these and other questions in the following article.

To understand what Class D is, we will first review the other four standardized classes of amplification in audio.

Class A is the traditional amplifier, a fully linear amplifier requiring a voltage supply identical to the entire dynamic swing demanded, that is, the amplitude of the amplified signal. That is why the power supply is particularly important, since it must be able to deliver at least a signal whose voltage is 200% of the maximum output signal. Although this method is still considered the purest amplification system today, it is rather inefficient. We speak of efficiency when relating the energy consumption demanded to the energy (or power) delivered at the output, a value expressed as a %. The efficiency of a Class A amplifier is barely 20% (that is, of 100% of the energy the device consumes, only 20% can be delivered at its output). The remaining 80% of the energy is converted into heat, so powerful heat sinks and even fans must be used.

Class B is more efficient than Class A, thanks to the use of a "push-pull" configuration. The result is similar to using two amplifiers, one for the positive part of a sine wave and the other for the negative part. The main problem is the error during the switchover between these two amplifiers, which results in what is called "crossover" distortion.

If we combine the best of the two previous classes, we arrive at Class AB amplifiers, a combination that efficiently solves the problem of Class A's low efficiency and noticeably reduces "crossover" distortion. In fact, it is the most widely used amplification system in home equipment, with Class A relegated to more audiophile solutions.

There is a fourth class, Class C, which, since it is used almost exclusively in radio-frequency transmission, we will leave aside in this article.

Class D

The "D" does not stand for digital at all. In fact, it is obvious that it makes no sense to take "digital amplifier" literally since, if we amplify a binary signal made up of 0s and 1s, what would we get? Bigger 1s and 0s? In reality, many people refer to Class D as digital amplification because the output devices are either working or not working (they are either on or off), an operation very similar to the binary concept (either 1 or 0). It also makes sense insofar as a binary (and therefore "digital") signal can be used as a guide or pattern to drive the amplifiers correctly.

The concept of the Class D switching amplifier dates from the 1950s. The first developments were tube switching amplifiers, in which the inherent limitations of tubes (such as their low current) reduced the amplifier's efficiency, limiting output power. In the late 1960s, bipolar transistors became the practical alternative to tubes, making it possible to improve switching amplifiers considerably, with higher efficiencies at low frequencies.

Broadly speaking, and to put it simply, Class D is based on the same principles as switch-mode power supplies, which we review in this same issue of Cine en Casa in the Face to Face section. The difference is that it is an audio signal that modulates the switching action.

Both switching solutions use pulse-width modulation (PWM) to control the on and off cycles of the power switching transistors. The result is a truly efficient amplification system, thanks to minimal voltage loss when switching between transistors during conduction or, in other words, almost all the energy is transferred directly to the load. During the off period there is essentially no current flow.

For audio applications, a very high frequency is used (at least four or five times the 20,000 Hz bandwidth of the audible sound spectrum), whose pulses vary in width according to the amplitude of the signal to be amplified. The input signal, in analog form, is compared with a triangular signal whose period depends on the switching frequency and whose amplitude corresponds to the dynamics of the desired amplified signal. Some describe this conversion method as 1 Bit (Sharp has long used this amplification system under that name), but the important thing is to recognize that this conversion method is completely linear (retaining the best of Class A): the relationship between the audio amplitude and the variable-width pulses in this system is perfectly linear.

The signal resulting from this comparison drives the output of the MOSFET transistors.

Audio and Class D

The principles of Class D suggested that it was not a suitable candidate to replace Class AB amplifiers. Even so, the development of new, advanced power semiconductors has allowed Class D to be used even in professional sound applications. Today the signal-to-noise ratio values are impeccable, and they are linear enough to outperform Class A without difficulty (not to mention the popular Class AB).

The advantages of Class D relate almost exclusively to its high efficiency, which makes it possible to go from the practical 20% of Class AB to 75% in Class D. This efficiency, as we have already said, reduces heat emission, which makes it possible to work with devices that run at low temperatures. This reduces the need for heavy, large heat sinks, and at low power levels fans are unnecessary. It also reduces the need for complicated, heavy traditional power supplies, with the latest switch-mode power supplies used in many cases. In fact, it is worth remembering that Class D and switch-mode power supplies, although they share a certain technological resemblance, do not necessarily have to go together.

On the downside is the complicated design of the output filter. As we have said, the signal to be amplified is associated with a very high-frequency triangular signal. The output filter must be able to eliminate this frequency, and a typical high-pass filter is not sufficient (we are all aware of the importance of harmonics, for example). The first noteworthy point is the reduction of up to 3 dB in the amplifier's bandwidth, and the second is the final result after this control waveform has been removed. When designing this output filter, it is important to attenuate this carrier frequency as much as possible without affecting the audio band. Even so, residues almost always remain, which designers sometimes try to solve with drastic equalization in the uppermost band.

Class D in a diagram

Below we can see a basic diagram of what Class D means. The input signal (which in this case is an analog signal but could be a binary data stream) is used together with a triangular waveform to control two amplifiers, usually MOSFETs, which, after a low-pass filter whose job is to remove the very high-frequency triangular wave, generate the high-power signal that drives the loudspeakers.

The heart, or key part, of Class D is framed within the red box: the PWM modulator.