Amplificador Clase D

1 -   Introduction
A completely new technology for audio amplification has been evolving during the last 15-20 years that has a clear benefit over current widespread Class-A, and AB topologies. We are talking about the so-called "Class-D". This benefit is mainly its high power efficiency. Figure 1 shows typical efficiency curves vs. Output power for Class-B and Class-D designs.
The theoreticalmaximum efficiency of Class-D designs is 100%, and over 90% is attainable in practice. Note that this efficiency is high from very moderate power levels up to clipping, whereas the 78% maximum in Class-B is obtained at the onset of clipping. An efficiency of less than 50% is realised in practical use with music signals. The PWM amp's high power efficiency translates into less power consumption fora given output power but, more important, it reduces heatsink requirements drastically. Anyone who has built or seen a high-powered audio amplifier has noticed that big aluminium extrusions are needed to keep the electronics relatively cool. The loading on the power transformer is also reduced by a substantial amount, allowing the use of a smaller transformer for the same power output.

Figure1 - Efficiency Comparison for Class-D and Class-AB
These heatsinks account for an important part of the weight, cost and size of the equipment. As we go deeper in the details of this topology, we will notice that a well behaving (low distortion, full range) Class-D amplifier must operate at quite high frequencies, in the 100KHz to 1MHz range, needing very high speed power and signal devices. Thishas historically relegated this class to uses where full bandwidth is not required and higher distortion levels are tolerable - that is, subwoofer and industrial uses.
However, this has changed and thanks to today's faster switches, knowledge and the use of advanced feedback techniques it is possible to design very good performance Class-D amplifiers covering the whole audio band. These featurehigh power levels, small size and low distortion, comparable to that of good Class-AB designs. (From now on, I will refer to Class-A and AB topologies as "classical").

Complete 400W Full-Range Class-D Amplifier Module (Courtesy of ColdAmp)
From the DIY perspective, Class-D is rather unfortunate. Because of the extremely high switching speeds, a compact layout is essential, and SMD (surfacemount devices) are a requirement to get the performance needed. The stray capacitance and inductance of conventional through-hole components is such that it is almost impossible to make a PWM amplifier using these parts. Indeed, the vast majority of all ICs used for this application are available only in surface mount, and a look at any PWM amplifier reveals that conventional components are barelyused anywhere on the board. Since SMD parts are so hard to assemble by hand and the PCB design is so critical to final performance, DIY versions of PWM amps are very rare indeed (I don't know of any).

2 -   How Class-D Works
In classical amplifiers, at least one of the output devices (let them be bipolar transistors, MOSFETs or valves) is conducting at any given time. No problem so far, but theyare also carrying a given current where there is a voltage drop between collector-emitter / drain-source etc. Since P = V * I, they are dissipating power, and even if there is no output a small quantity of current must pass through the transistors to avoid crossover distortion, so some dissipation is present. As the output voltage increases, for given supply rails the voltage drop across thetransistors will fall, but the current increases. At saturation (clipping), VCE or VDS will be low, but current is quite high (Vout / Rspk). Conversely, at low power levels, current is small but voltage drop is large. This leads to a power dissipation curve that is not linear with output power. There is a non-zero minimum dissipation (zero percent efficiency), and a point where maximum efficiency is...