Class D amplifiers have been used in designs where audio performance and EMI are sacrificed for increased efficiency. Today, however, several new techniques have lead to the introduction of ultra-low EMI, high performance Class D amplifiers. The following article describes how system designers can benefit from these new technologies.
As portable, battery-operated devices proliferate; Class D amplifiers continue to gain popularity, due to their inherent efficiency advantages. Most Class D systems now operate above 80% efficiency. In the past, engineers have had to sacrifice audio performance, and increase board space and system cost in order to realize these efficiency gains. Recent advances in Class D technology have addressed the short comings of past architectures, simplifying system design, and reducing solution cost.
Large filters, electromagnetic interference (EMI), or RF interference (RFI) and poor Total Harmonic Distortion + Noise (THD+N) are issues that commonly accompany Class D amplifiers. Recent architectures use the inductive nature of the loudspeaker itself to extract the audio component from the pulse width modulation (PWM) square wave output, eliminating the output filter for audio purposes. However, removing the filter has increased the EMI emitted by this filterless topology. The newest devices have been able to minimizing EMI and improve THD+N, without sacrificing efficiency, as described below.
EMI is important to many designers as it can interfere with ICs and electronic devices elsewhere in the system. Engineers are also challenged with compliance to the standards set forth by governing bodies such as the FCC, CE, Mil-Std-461, and proprietary automotive. The first EMI suppression feature semiconductor suppliers implemented is spread spectrum modulation. Spread spectrum modulation differs from traditional PWM as the switching frequency of the output bridge changes in a band around a center switching frequency. The center frequency, frequency spread and frequency variation method can differ from device to device, but as long as the frequency variation is random, the peak radiated energy will decrease. This is due to the fact that the electromagnetic energy is spread over a wider frequency band. The total high frequency energy remains the same compared to a fixed frequency device, but the peak noise at any one frequency is reduced.
Figure 1 demonstrates near field EMI measurement for a fixed frequency device vs. a spread spectrum device. As the red lines indicate, peak energy is reduced.
Spread spectrum is effective when implemented correctly, and does not adversely affect efficiency or THD. It is now implemented in several devices.
To further reduce the EMI profile of a device, semiconductor manufacturers have implemented edge rate control (ERC). The high frequency energy of a Class D output is contained in the edges of the square wave output. The faster the output rise or fall time, the more high frequency energy the edge contains. Therefore, if the output transition time can be decreased, the amount of high frequency energy released by the system will be reduced.
If not taken into account, decreasing the transition time can also have negative effects on Class D performance. As more time is spent in the active region between states, the output devices dissipate more power, decreasing efficiency. Reducing the rise and fall times also results in a PWM signal that deviates from a perfect square wave, introducing error into the reproduced audio signal and increasing THD+N.
Although edge rate control potentially has a negative impact on the overall performance of a Class D amplifier, the EMI improvement has compelled designers to make marked improvements to ERC technology. When properly implemented, efficiency losses and added THD+N can be minimized.
Devices implement an available enhanced emissions suppression (ES) system. E2S improves efficiency by slowing the output transition time for only a portion of the edge. In this way, EMI is minimized but power dissipation is also reduced to a level on par with non-ERC Class D amplifiers. PWM audio signal errors introduced by the ERC are corrected by an internal feedback loop, reducing THD+N and improving audio quality.
Performance is shown in Figure 2. The device passed the FCC Class B standard, completely unfiltered, driving a speaker with twenty inches of unshielded, twisted pair cable. Figure 3 shows the same test with a spread spectrum-only device. As illustrated, the E2S Class D amplifier achieved extremely low EMI, while maintaining audio performance.