Class D amplifier has been greatly developed in the past few generations of products. The system designer has greatly improved the durability of the system and its audio quality. In fact, for most applications, the benefits of using these amplifiers have far exceeded their disadvantages.

    In the traditional Class D amplifier, the analog or digital audio signal is converted into PWM signal by the controller before being amplified by the power MOSFET integrated into the power back-end device. These amplifiers are highly efficient, use very small heat sinks or do not need heat sinks at all, and reduce the requirements for power output power. However, compared with traditional class A/B amplifiers, they also have inherent problems in cost, performance and EMI. Solving these problems is the new trend of class D amplifiers.

    Since the birth of Class D amplifier, a large number of radiated EMI caused by its own rail-to-rail power supply switch characteristics has been puzzling the system designers, which will make the equipment unable to pass the FCC and CISPR certification.

    In the class D modulator, the digital audio signal is converted into a PWM signal by comparing the audio signal with the high-frequency fixed frequency signal and modulating the result on the carrier of the fixed frequency. The generated signal is a fixed carrier frequency with variable pulse width (usually several hundred kHz), and then these PWM signals are amplified by high-voltage power MOSFET. The amplified PWM signal is then removed by low-pass filter to recover the original baseband audio signal.

    Although this topology is very effective, it also leads to some undesirable consequences, such as a large number of radiated EMI. Because the modulator uses a fixed frequency carrier, it will generate multiple harmonic radiation of the base carrier. Moreover, due to the switching characteristics of the PWM signal itself, overshoot/undershoot and ringing will generate a fixed ratio of high frequency (in the range of 10 to 100 MHz) radiated EMI. In order to suppress radiated EMI, the development trend of the latest generation of PWM modulator is to use spread spectrum modulation technology.

    The spread spectrum modulation technology is used to expand the spectrum energy of the switching PWM signal in a larger bandwidth without changing the content of the original audio. An effective way to improve the high radiation EMI of the traditional modulator is to change the two edges of the PWM switch signal, as shown in Figure 1. The signal is centered on the carrier frequency, but any edge is not repeated periodically. This not only maintains the fixed carrier frequency, but also greatly reduces the radiation energy on the carrier frequency because the edge does not jump at a fixed rate.

    Compared with class A/B amplifier with excellent performance, class D amplifier has poor audio performance, not only large distortion, but also narrow dynamic range. Therefore, the current designers of Class D amplifiers must improve their performance. By integrating high-performance sampling rate converter (SRC) and Δ-Σ The processing technology and the new generation of solutions have greatly improved the distortion (THD+N), and the dynamic range has also exceeded 100dB.

    At present, one noise source of Class D amplifier is the jitter of audio sampling clock. The clock is usually generated by SOC (MPEG decoder, DSP, etc.). Even a small jitter can quickly affect the performance of conventional Class D amplifier, because the audio clock is associated with the output clock of the modulator.

    One way to solve this problem is to use SRC technology. Because SRC uses a locally stable clock source to synchronize the clock of digital audio, such as a quartz crystal oscillator, the output jitter of the modulator is actually independent and unrelated to other audio clocks. Another advantage of SRC is that no matter how the sampling rate of input audio fluctuates, its output switch ratio is fixed, which is different from PLL-based modulator. When the audio input source changes or the input clock is missing, SRC also improves the durability of the system by eliminating audible noise.

    Similar to the technology used by the current high-end DAC, through the integration of high-level Δ-Σ Processing technology, audio quality of Class D amplifier has also been improved. be based on Δ-Σ The modulator of the technology adopts internal feedback that can reduce the modulation error. By reducing the sampling error, the modulator can improve the output distortion and obtain better sound quality.

    In order to reduce the cost of Class D amplifier, the designer adopts a half-bridge amplifier topology at the power amplifier stage to reduce complexity and material cost. Because the output of half-bridge structure is usually half of the full bridge, the number of power MOSFETs and external filter devices will be reduced by half. This also increases the number of power channels per unit of back-end devices. However, the half-bridge amplifier also needs a direct-isolating capacitor at the output end, and is extremely sensitive to the noise on the power supply line.

    During startup, the DC isolating capacitor must be charged to the bias point (half of the voltage of the high-voltage power supply mains). If the output signal does not rise from the ground potential to the offset point, there will be a loud "popping" sound in the speaker (boot impact sound). The new class D amplifier uses a pre-charged capacitor to keep the speaker silent when starting.

    One of the ways to keep the speaker without impact sound when charging the direct-isolated capacitor is to use digital voltage boost technology, that is, to slowly increase the PWM duty cycle from the non-switched state to 50%. This will not produce a large "pop" sound in the speaker, but because of the large amount of transient current generated during MOSFET switching, the speaker is not soundless.

    Another way to keep the speaker without impact sound when charging the direct-isolated capacitor is to use analog voltage boost technology. In this type of voltage rise, a current source charges the capacitor to the bias point. Once the voltage at both ends of the capacitor reaches the bias point, the current source will be turned off.

    Since the half-bridge is a single-ended topology, there is no common mode suppression in the differential full-bridge topology. In a full-bridge amplifier, since the differential output of the amplifier is powered by the same voltage source, the noise on the common voltage source will be offset at the output end. In the half-bridge topology, any AC ripple noise on the amplifier power supply will be directly coupled to the output. Because the half-bridge topology is sensitive to power supply noise, it is often necessary to provide power supply rejection feedback (PSR) circuit to reduce noise.

    Analog class D amplifier has many inherent PSR characteristics, while complete digital class D amplifier does not. In the current digital PSR scheme, an external ADC is usually used to monitor the power supply of the amplifier.

    The feedback and noise cancellation processing is performed in the digital domain of the modulator. Some manufacturers only use this feedback method to compensate for the impact of AC noise coupled from the power supply mains to the PWM output that reduces the system performance. Other manufacturers also use it to compensate for the change of DC power supply voltage (voltage drop) caused by load change, for example, the fast surge current required by the bass unit (subwoofer), or the voltage fluctuation of the power supply line. The advantages brought by PSR feedback in AC and DC devices have been extended to full-bridge amplifiers, and improved the isolation between channels in current multi-channel home theater amplifiers, effectively eliminating crosstalk and line voltage changes before reaching the output.