DesignWare Technical Bulletin Article

Search Tools


Advanced Audio Drivers - The Rising of a New Class of Drivers

By Joćo Risques, Product Marketing Manager, Synopsys Inc.

Analog audio processing is essential to consumer electronics. Traditional Class-A/AB driving architectures are becoming inefficient to address today's product requirements in terms of lower power and reduced cost for better performance. In the recent years, Synopsys took the lead by making available highly efficient and very low area advanced audio driver IP (intellectual property) that allows embedding Class-D and other advanced audio drivers on SoC.

Class-D Drivers - Making the Best Out of Each Milliwatt
Class-D digital audio drivers - power amplifiers - offer significant power savings over traditional Class-AB analog amplifiers. While Class-AB amplifiers are at best approximately 60% efficient for near full power outputs, Class-D amplifiers achieve 85% to 95% efficiency. For driving a 0.5 W or 1 W loudspeaker, this represents significant power savings.

Class-D audio drivers were historically used in bulky high power electronic devices. With the introduction of area optimized, power efficient and increasingly high quality sound characteristics - low power drivers IP, the Class-D audio architecture also offers great benefits for portable devices featuring loudspeakers (mobile phones with hands free functions, portable gaming consoles, etc.).

Until late 2007, Class-D audio drivers required a separate IC. With the introduction of the first embedded Class-D driver IP by Synopsys, it is now possible to integrate Class-D audio drivers into SoC designs. Power efficiency is over 85%, peaking at a remarkable 94% efficiency while delivering Hi-Fi audio quality at 90 dB dynamic range and 0.05% harmonic distortion at 1 W output power (Figure 1). And it occupies just 0.4 mm2 in a 65 nm technology.

With Class-D Audio power amplifiers IP available, today's system designers have audio capabilities for the creation of low-power, low-cost custom SoCs.

Figure 1- Efficiency (left) and Performance Test Results of Synopsys 1-Watt Class-D driver

Linearized Digital Audio - Making the Best Out of Each Square Micron
Digital Audio playback systems, i.e., audio playback channels with line driving capability built using digital standard cells only, are a solution to reduce area (cost) when compared with the traditional DAC plus line driver architectures and has been available for quite some time. Arguably, digital audio solutions offer a number of benefits, namely the smaller area (especially in deep sub micron) and lower risk (digital designs are more predictable than mixed signal designs in terms of production yields).

Despite the benefits above, digital audio solutions have consecutively failed to make traction in the highly competitive audio arena where the traditional mixed signal converter plus driver solution is still preferred. The main reason is that harmonic distortion for such an architecture depends mainly on the digital output pad's buffer characteristics (driving capability, rise and fall times matching, etc.) and has been proven to be quite low (0.5% at best). For most, if not all applications requiring audio playback capability, this is far from adequate (Hi-Fi systems require better than 0.005% harmonic distortion while some lower end applications require at least better than 0.05%).

By introducing the linearized digital audio architecture back in 2008, Synopsys took the lead by making available a solution that combines the best of two worlds: the very compact area offered by the digital audio architectures and the added linearity provided by the linearization architectures developed for our Class-D products. The result is a product that saves up to 50% in silicon area while delivering the same Hi-Fi audio quality at 96 dB dynamic range and 0.005% (-86 dB) harmonic distortion.

This product is particularly suited for home entertainment applications (DTV, STB, etc.) where multiple channels are needed (5.1 or 7.1 arrangements) and where the consequent area savings are more relevant. Figure 2 shows the layout of a 6 channel (5.1) Linearized Digital Audio product with 1.2sqmm analog area only.

Figure 2- Mask layout of a 6-channel Linearized Digital Audio product

Class-G - Expanding Battery Life
In the mobile world, battery life is the most relevant differentiator and the 100-hour-playback-time is the holy grail which all modern products are targeting.

Traditionally, headset drivers have been designed using Class-AB amplifiers which achieve a moderate harmonic distortion performance (0.02%) in a very compact layout. With the increasing requirement to reduce power consumption and increase power efficiency in order to extend battery life, the Class-AB driver's power efficiency limitations (maximum 65% efficiency for nearly full power output) become more relevant than ever before. Though many audio products enable headset drivers with output power reaching above 50 mW (over 16 Ohm), the fact is that the comfortable power levels are significantly lower - on average, a 2 mW audio signal is perceived as very "loud" by the listener. And for 1 mW output level, a traditional 3.3 V supplied Class-AB driver the efficiency comes all the way down to less than 10%.

The main reason for the low efficiency at comfortable audible sound levels is the difference between the supply (3.3 V) and the output level (360 mVpp for 16 Ohm loads). Obviously a driver capable of operating at a half the original supply would immediately increase the efficiency - and it is possible to design such a driver. However, such an approach would clip the audio signal whenever strong inputs were present - in audio clips, it is common to have a wide range of signal strengths, from the weak whisper to the loud door slam - designing for the weaker signals and improving the efficiency clips the louder signals; conversely while designing for the louder signals degrades the efficiency in the presence of weaker signals.

The solution is to develop a product where the supply can change as a function of the signal strength - also known as the Class-G operation. Such driver's supply can be kept to a minimum for weak signals thus improving the efficiency and increased whenever stronger signals must be reproduced.

Figure 3- Class-AB (left and center) versus Class-G operation

Figure 3 illustrates the advantages of the Class-G operation. Class-G drivers are suitable for mobile applications featuring Hi-Fi Audio Playback requiring everlasting battery life.

To Each Their Own
While Class-A/AB drivers still play a significant role in today's Audio Playback systems, mainly because of its still prominent advantages like compactness, high linearity and most notably very high flexibility, market trend to require added low power consumption and high power efficiency together with ever more lower area have been driving the need for more advanced types of audio drivers.

Synopsys advanced audio drivers range addresses multiple different application requirements. While Class-D drivers are suitable for any mobile applications with built in loudspeakers, Linearized Digital Audio IP is suitable for multichannel home entertainment line drivers and Class-G drivers are particularly suitable for applications requiring driving headsets; all of them delivering Hi-Fi Audio quality.

Advanced audio drivers have traditionally been only available as standalone ICs. However, it is today possible to embed Synopsys advanced audio drivers IP on a SoC thus considerably contributing for its overall increased power performance and lower area, and ultimately its competitive differentiation.