When the world's first commercially available audio product using a digital amplifier appeared in 1998, it created quite a stir with its performance and ease of use.
Digital amplifiers provide several advantages over the traditional analog-based circuit, such as fine detail, noise immunity, high efficiency and the configurability that the digital system allows. In addition, because audio-processing design is now software-centric, changes can be made quicker with less impact to the design schedule. As a result of these advantages, the digital amplifier has come a long way in just a few short years and is predicted to become the dominant audio amplifier technology in the market within five to 10 years.
The digital amplifier has only two output states. Using power Mosfets as on-off switches, it applies a fixed dc voltage across a speaker in either a positive or negative direction. This causes either a positive or negative current to flow into the speaker. The digital amplifier re-creates the audio signal by adjusting the duty cycle of this positive/negative signal at a high rate (typically 384 kHz). To generate a positive current into the speaker, the duty cycle must be greater than 50 percent. Negative current requires less than 50 percent and at 50 percent the current is zero.
The speaker can be controlled very precisely in this manner and the digital amp is therefore capable of generating very fine detail, essentially creating an exact reproduction of the source material. Other advantages include the topology's high efficiency (greater than 90 percent) and the virtual noise immunity of the digital signal path.
Some in the industry also call analog-input Class-D amplifiers "digital amplifiers" because they also have only two output states. The analog-input version is generally used in systems where high efficiency is required (cars, cell phones, laptops, etc.), but where the audio comes from an analog source. In this article, "digital amplifier" will refer to digital input only.
Two audio pathways
The digital amplifier system keeps the audio signal in the digital domain virtually all the way to the speaker, while the analog-based system converts the signal to analog very early in the process.
The digital amplifier consists of two devices, the pulse width modulation (PWM) processor and the power stage. The PWM processor converts the pulse code modulation signal to PWM in addition to performing 48-bit digital processing, such as basic treble, bass, volume control and more advanced functions like dynamic range compression or loudness. In contrast, the same functions are usually implemented after the digital-to-analog converter in an analog system, typically with op amp-based ICs.
Digital domain processing offers many advantages to the audio design engineer. For example, it is much easier to change configurations or functionality digitally and this can even be done on the production line or in the end user's home. The 48-bit processing ensures that artifacts from the processing remain far below the noise floor.
Both digital and analog-based systems contain an interpolation filter (IF), which oversamples the original signal. Following the IF, the digital amplifier performs fifth-order noise shaping, which pushes noise from the conversion process out to ultrasonic frequencies, leaving the audio band virtually free from noise. The signal is then converted to PWM and since this conversion is nonlinear, correction must be applied at this point.
The PWM signal is a one-bit (i.e., two-level) signal that is sampled at an extremely high rate (4,096 times the input sample frequency or about 200 MHz). Timing is critical with PWM to re-create the input signal and so internal clocks with extremely low jitter are essential. Analog-input Class-D amplifiers generate the PWM signal through advanced analog circuitry.
Once the signal is converted to PWM, it then passes to the power stage where it is checked for error conditions and timing control is applied. The signal is then passed to gate drivers, which control the Mosfet switches. Other functions in the power stage include detection of and recovery from error conditions such as overcurrent, overtemperature and under-voltage. The analog-input Class-D amplifier has a similar power stage architecture.
At the last step, a low-pass inductor-capacitor filter after the power stage removes high-frequency components from the audio signal to reduce electromagnetic interference, and this is effectively the digital-to-analog conversion. Without this inductor and capacitor filter the speaker itself becomes the D/A converter.
This analog conversion introduces no distortion to the signal, because the digital amplifier converts the signal to analog with passive components. The conversion from digital to analog occurs at the final output voltage, so it is unlikely to pick up noise that will be audible. In contrast, the analog conversion occurs as the first step in an analog amplifier and is at a low-level voltage, which is then processed and amplified. Any noise on this low-level signal from processing or from coupling also will be amplified and will affect the final signal quality.
Kevin Belnap (firstname.lastname@example.org) is digital audio product-marketing manager at Texas Instruments Inc. (Dallas).