Class D amplifiers continue to evolve as a result of device technology advancements

Class D is redefining the trade-offs in audio power amplifier design, simultaneously delivering smaller size and higher power density with better sound than Class AB audio systems. Class D also has the advantage of offering low distortion and more stable dynamic response compared to Class AB audio systems while power semiconductors also help boost efficiency and facilitate audio performance.

Recap on class D operation
As figure 1 illustrates, the class-D amplifier converts the input audio signal into a series of pulses which have instantaneous average value proportional to the input signal. This binary signal switches the power MOSFET, creating an amplified version of the PWM. A passive low-pass filter attached to the PWM switching stage removes high-frequency components to recover the amplified audio signal.

Class D under the skin
The inherently high efficiency of a class-D amplifier allows parts such as heat sinks and cables to be downsized enabling audio equipment to be far smaller than has historically been possible. This not only helps improve aspects such as styling and portability, but also contributes to improved audio performance. Every current loop inside an amplifier forms magnetic couplings with other current loops, effectively creating many small transformers that can interfere with nearby circuits and components. When the amplifier is physically smaller, the magnetic flux due to these current loops is reduced. The susceptibility to flux is also lower. In practice, the impact of magnetic flux is inversely proportional to the square of size. A switching transition creates high frequency noise from parasitic impedance along the switching circuit. To achieve high fidelity of output, faster and cleaner switching waveform is essential. Lowering parasitic impedance carries out faster and cleaner switching transitions. Putting all necessary circuits in a small footprint is inevitable and a newer Class D solution provides that opportunity.

The challenge in linear topology is that these semiconductors are current output devices, meaning that output impedance is very high (>>10k ohms) when designing an amplifier as a voltage output with very low output impedance (<<1 ohm). A voltage negative feedback is the only option makes the low output impedance occur. However, if there is a back EMF coming back from the speaker, the output voltage is left high impedance until the feedback loop realized that there is an error occurred in the output node. This means making every effort of maintaining output voltage falls behind.

A Class D amplifier today still requires support for negative feedback for different reasons such as influences from inserted blanking time in power stage and power supply voltage fluctuations.

The good news is that the Class D amplifier is a voltage output topology before applying a single feedback loop. The output stage shunts out back EMF and incoming noise from the speaker at the instance of injection so that the feedback signal feeding back to the error amplifier is significantly less contaminated by the noise, making the control loop extremely stable under dynamic environment.

Compared to class AB, class D exhibits inherently superior response to changes in power demand, resulting in highly dynamic sound characteristics. This superiority comes from the completely different manner in which the class-D amplifier controls its output power. Even when the amplifier is idling, the output MOSFETs are turned fully ON or OFF alternately every 1.2µs. Hence the timing of the switching controls the power flow; to change from zero output power at idle to a half rated power output takes only 0.6µs until the next switching event is triggered. Hence the switching timing regulates output power, with no additional energy needed to commence the event. In other words the output power is controlled independently of the current or voltage demanded by the speaker, and is simply determined by when the PWM signal switches from one state to another. This ensures robust control of the power output to the speakers, and reduces distortion caused by the speakers’ back EMF.

In addition, there is no temperature-sensitive operating bias point in a class-D power stage. This enables the amplifier to maintain stable operation under dynamic power changes. The gain of the class-D stage is simply a function of supply voltage and a ratio of the ON time between the output MOSFETs, and is not related to temperature. One big challenge in class AB design used to be how to maintain the stability of the bias current under dynamic loading conditions, since the transistor gain when used in the linear region is strongly dependent on bias current, which in turn is highly sensitive to temperature. Class D eliminates this interdependency.

Key semiconductor components
Although class D allows a theoretically perfect amplifier, with 0% distortion and 100% efficiency, the actual performance is dependent on the quality of the components used. In particular, the output power MOSFETs and the controller IC each have a major influence.

A perfect MOSFET would allow the amplifier to have 100% efficiency resulting in zero heat generation. Today’s state-of-the-art MOSFETs allow a practical Class D to achieve above 90% power efficiency. Examining the On resistance (RDS(ON)) x gate charge (Qg) figure of merit (FOM) illustrates how closely the performance of today’s MOSFETs now approaches that of an ideal power switching device.

Consider the evolution of 200V rated MOSFETs over the last couple of decades. To achieve the highest efficiency in a Class-D amplifier, the conduction loss from RDS(ON) and switching loss dictated by gate charge Qg should both be as low as possible. The IRF640 from the 1980s, which has a planar structure, has RDS(ON) of 180mΩ with Qg of 70nC. The latest trench structure MOSFET IRFB4227 has RDS(ON) of 20mΩ with the same 70nC Qg. The FOMs has been reduced from 12,600 to 1400; a nine-fold improvement.

Control ICs
For the controller IC, effective noise isolation is a key requirement. IR has paid special attention to this aspect in the design of its class-D amplifier controllers, and uses proprietary techniques to prevent switching noise being coupled into noise-sensitive error-compensation circuitry. This has enabled the built-in noise-sensitive analogue error amplifier in the IRS2092 controller, for example, to provide a clean low-noise output despite the fact that the other side of the silicon chip just 1mm away is switching between -60V and +60V at very fast transition speed. This error amplifier is also well isolated from the switching of the external output power MOSFET.

To achieve low distortion, precise timing control of the PWM gate-drive signal is as crucial as ensuring precision analogue signal processing in the error amplifier section. A stable dead-time control and jitter-less level shifting is also needed to control the gate signals to the output MOSFETs. IR has developed special circuits that attain very low jitter in the signal path.

The IRS2092 combines these noise-isolation and timing-control features with robust protection circuitry to create a class-D amplifier that surpasses the audio performance possible with a conventional class AB approach in many aspects.

Integrated Class-D amplifier
Building on the same MOSFET and controller IC construction, IR’s PowIRaudio IR43xx ICs combine both components in a small surface-mount package.

Instead of integrating the controller and output power MOSFETs on a single piece of silicon, the IR43xx family features separate controller IC and MOSFET chips. This approach takes full advantage of the wide operating voltage range and strong noise immunity of the IRS2092, as well as the latest-generation application-optimized MOSFETs, to deliver the high efficiency and high audio quality achievable with class D within an even smaller footprint. Evolution is set to continue further, for example using new semiconductor materials such as Gallium Nitride (GaN) which can potentially bring a ten-fold performance increase.

The PowIRaudio family comprises the IR4301M, IR4311M, IR4321, IR4302M, IR4312M and IR4322, supporting full-bridge and half-bridge topologies from 20W to 320W per channel.

Class D offers smaller size and higher power density with better audio performance, and generates much less heat as the topology does not use temperature sensitive characteristics of power devices, making it robust. The lower heat dissipation allows for a smaller footprint that reduces magnetic coupling interferences.

The power stage is the key factor governing the performance of a class-D amplifier. The Power stage component holds the key to achieving great power and sound, taking advantage of the exceptionally high linearity and energy efficiency inherent in the class-D operating principle. Newer device technology moves these amplifiers closer towards the ideal by increasing efficiency and improving distortion.