Author: Infineon technology – Jun Honda, lead Principal Engineer for class D audio and Pawan Garg, system application engineer
Audio is a complex application, especially for the needs of all levels in the field of enthusiasts. The most high-end audio devices are usually expensive. Different types of audio amplifiers attract many users, who believe that their choices can best reproduce the true meaning of the original recording. Although there are many discussions on the advantages and disadvantages of various amplifier designs in various enthusiast forums, energy efficiency plays a very important role in many application fields.
Class D amplifier was first proposed in the 1950s (see Figure 1). Its main competitive technologies include class A, class AB and class B amplifiers, which use transistors in the linear region to reproduce the amplified version of the input signal as accurately as possible. However, the theoretical efficiency limit of these designs is less than 80%, and the actual efficiency is less than 65%. For class D design, the input signal is used to control push / pull switches with pulse width modulation (PWM) to allow them to operate in on and off modes. As a result, they will not work in their linear region, allowing the design to provide theoretically 100% efficiency and zero distortion.
Figure 1: basic block diagram of class D amplifier design.
In the early stage of development, until the emergence of silicon MOSFET with appropriate device parameters, there were no devices available in the industry to realize the full potential of class D technology. But since then, class D amplifiers have achieved great success, especially for battery powered devices such as smart phones, hearing aids and Bluetooth headsets, in which high efficiency and low heat dissipation are very favorable features. Of course, higher power amplifiers used in areas such as TVs and cars also benefit from class D technology, which makes compact designs require little or no radiator.
Recently, high electron mobility transistor (HEMT) specifications based on GaN technology have paved the way for better utilization of class D amplifier performance.
The new switch technology meets the requirements of class D amplifier
Class D amplifier can provide high efficiency and low distortion capability, which depends on the switching device selected. First, the on resistance must be as low as possible to reduce the I2R loss. Secondly, in order to support higher switching frequency, the switching loss must be minimized. Due to the loss of power devices, the efficiency of all kinds of amplifiers is usually very poor at the lowest power output. Only when the power output reaches a certain level, the efficiency begins to improve.
Class D amplifier can realize a so-called multilevel technology, in which the maximum output voltage is limited when the audio is output at a lower volume. This method helps to improve the efficiency at a low power output. As the audio goes to a higher output level, the entire voltage swing is available for switching devices. Zero voltage switching (ZVS) is used at lower output level, while hard switching is used at higher output level. Both modes of operation affect the loss of the switch.
In ZVS mode, the output is changed by inductor current commutation. Therefore, any switching loss in the switching device and the resulting power loss can be eliminated. However, in order to avoid shot through between the two devices, a small blank delay must be added to ensure that the off state of the previous switching cycle can be maintained before entering the on state of the next switching cycle. This will make the output waveform different from the expected waveform of PWM output, resulting in audio signal distortion. The blank delay time depends on the output capacitance of the power device used. Compared with Si MOSFET, the cost of Gan transistor is significantly lower, which means that the blank delay time can be kept to a minimum, thus minimizing the distortion.
Hard switching at high power output means that the output voltage is nonzero when the power device is on or off. The Si MOSFET has a diode which accumulates reverse recovery charge (qrr) after the switch is turned off. Before the opposite switch enters the on state, it needs to be discharged, which takes some time. Gan transistors are very different here, because there is no internal body diode, so there is no qrr. This results in clearer switching waveform, improved distortion coefficient and higher overall efficiency.
Unfortunately, when we use Gan technology, we also need to deal with the challenges brought by Coss. However, the energy of Gan is significantly lower than that of Si MOSFET, resulting in less energy dissipation in the next conduction cycle. Because this has a great influence on the high frequency loss, Gan shows a very beneficial improvement compared with Si. Most importantly, switching to Gan technology can also provide lower on resistance in a smaller die size, which enables engineers to achieve more intensive and compact audio solutions in addition to providing better audio quality.
How to embody the advantages of Gan in design
Like similar silicon devices, GaN HEMT devices also have gate, drain and source terminal connections. Two dimensional electron gas (2DEG) layer can provide an electron pool, which can realize short circuit between source and drain with very low resistance. When no gate bias is applied (VGS = 0V), the p-GaN gate stops conducting. However, it should be noted that Gan HEMTs are different from Si MOSFETs in that they are bidirectional and reverse current flow is allowed if the drain voltage becomes lower than the source voltage. The absence of bulk diode also greatly eliminates the common PN junction related switching noise in Si MOSFET, thus providing a more “clean” switch (see Figure 2).
A 250W device based on class D technology is igt40r070d1 e8220, which can provide 70m Ω RDS (on) (max) and 200V class D driver IC (irs20957s), and can provide 160W power for 8 Ω load without radiator (see Figure 3). At 100W, thd + n is only 0.008%. When the switch is set to 500KHz, thd + n measurement shows that when the amplifier changes from ZVS to hard switching area (the power is only a few watts), there is no obvious change in distortion, and the hard switching area is very clean with little noise.
Seventy years ago, the introduction of the Class-D amplifier concept provided unprecedented but theoretically reasonable audio fidelity and excellent efficiency. Although the performance of traditional silicon MOSFET has been greatly improved and the design has made continuous progress, the influence of qrr and coss limits the higher switching frequency, limits the further improvement of efficiency, and eventually leads to audio distortion in class D design. To achieve a lower RDS (on), a larger chip size is required, which also means that a larger size is needed to achieve a more energy-efficient design. With the application of Gan transistor technology, the qrr is eliminated, and the cost is greatly reduced. It can provide the best thd + n results and operate at a higher switching frequency. The low RDS (on) (max) inherent in the compact package enables class D Gan amplifiers to provide high frequency fidelity in a small volume without cumbersome heat dissipation solutions.