Design hi-fi audio into a handset

The entire mobile services supply chain has been facing falling ARPU throughout the 2/2.5G era. Their response has been to vigorously encourage subscribers to consume a wider variety of services, in addition to basic telephony, using mobile handsets.

Fortunately, 3G bandwidth now makes many high-value services feasible, allowing service providers to generate revenue from activities such as music file sharing, gaming, and streaming video. But consolidating all the necessary capabilities into the most popular handset form factors brings significant challenges in manipulating numerous audio signals, and managing noise and power throughout the device. To allow these new services to become successful also requires handset designers to achieve a quantum leap in audio playback performance.

More sophisticated audio functions are a basic requirement, if emerging multimedia and music-oriented handsets are to deliver a persuasive user experience to stimulate the growth that mobile network operators are looking for. At the very least, louder and more powerful speaker drivers are a must, to deliver better speakerphone operation. High-fidelity (hi-fi) audio is also necessary through headphones, for personal music, as well as during voice calls for added value features such as karaoke. However, the small enclosure size of the typical mobile handset doesn’t predispose the device to hi-fi audio performance.

As far as improving speakerphones is concerned, the latest high-output speakers deliver much improved audio quality. But their high supply voltage requirements conflict with the generally falling operating voltages elsewhere in the handset. Designers therefore must ensure a stable supply near battery voltage level, in addition to digital logic and analog rails.

Manipulating multiple streams
Combining the numerous telephony features of a modern handset with advanced music features results in numerous audio streams comprising various digital formats as well as analog. The digital signals are also present in a range of sample rates and word lengths. Arranging for these varied signals to coexist without interference is demanding enough. But extensive, flexible mixing capabilities are also required, to support advanced entertainment features, such as music playback during voice calls.

Multimedia handset design also calls for flexible digital-audio sampling and much more powerful processing capabilities to enable hi-fi quality audio recording and playback features. Sophisticated audio processing such as multi-band equalization or 3D sound processing, for example, will become necessary to optimize the signal for playback through a speakerphone or headphones. By tailoring filters to match the human ear’s response when listening through headphones or on speakerphone, for instance, designers can compensate for some of the shortcomings imposed by the small handset enclosure and relatively limited speaker response. A sophisticated equalizer can also enable a range of audio effects, including adopting settings optimized to give the most authentic sound for certain music types. These settings may need to be invoked in one or more ways, such as by user-defined preferences or manual switches, or automatically by the handset software.

Power management
Despite the additional functionality of multimedia phones, users won’t accept shorter talk and standby times. Because the RF stages will continue to account for most of the power consumed, sophisticated audio can be implemented with little impact on power consumption, especially if a good power-management strategy is devised.

Careful design and operation of signal paths, for example, allows tight control over portions of the circuit to be turned off when not required. For example, because voice memo operation only requires basic mono audio, it can be run with just one ADC active, thereby saving the power consumed by two DACs and the DSP in the stereo audio signal chain. Reducing sample rates during voice calls can save power without suffering any noticeable loss of voice quality. A more sophisticated power-saving techniques is to reduce the analog bias levels to achieve an optimal balance between audio output and power consumption.

Integrated audio hub
As discussed, handset designers must meet the diverse challenges presented by these speaker drive, signal processing, noise performance, and power management demands within a small physical footprint and within the tight design cycles that characterize the mobile market. Throughout the 2/2.5G era, handsets designers usually relied on an integrated baseband chip set featuring a mixed-signal voice codec analog front-end that typically had no greater audio capability than necessary to support basic voice telephony. This would be coupled using discrete interfaces to a low-power DSP performing the channel coding and compression/decompression algorithms to lower the data rate over the air interface.

Clearly, emerging multimedia phones must pack a bigger punch in terms of codec capabilities to support the multitude of new services that rely on high-quality audio performance. Applying mixed-signal design to combine ADC, more powerful DSP, audio mixers, flexible speaker outputs, and power management for sophisticated functions onto the same silicon, effectively creating a hub for all audio signal paths, will enable a turnkey solution to reduce development time, risk, footprint, and bill of materials.

An integrated solution such as this must be conceived from an application-centric perspective, to facilitate enhanced support for the growing number of use scenarios. In the future, these are likely to include audio playback during calls to allow the called party to hear playback as well as the caller’s voice. Users will also likely want to initiate normal mobile phone functions while listening to MP3s or FM radio. Other functions becoming attractive to music phone users include karaoke during voice calls, with full mixing capability for headphone playback, as well as recording in I2S format, facilitating sharing of karaoke files over MMS or email.

To support these emerging use modes, such a device will at the minimum require flexible digital and analog interfaces to allow mixing of synthetic content, such as ringtones or midi files with audio or voice. These diverse data streams also have different sample rates, which adds to the complexity of the challenge.

Integrating saves power
Bringing the multitude of signal paths necessary to support flexible multimedia features together into a single-chip audio hub facilitates power management (see the figure). An audio codec with integrated DSP, for example, makes it possible for the handset to play MP3 music or MPEG4 movie playback with only the audio codec and main multimedia processor turned on. In addition, automatically reducing signal levels such as the digital voltage swing in the ringtone signal path, or analog bias levels during voice calls is another way of unlocking power savings at no audible cost in terms of ringtone playback or voice quality.

High-quality audio on a handset could be an attractive selling point. Shown are the pieces that are required.

A convenient way of invoking these modes is also required, for example, by ordinary register operations within software. In this way, many of these modes can be made to operate transparently to the user.

Maximizing quality
The concept of the multimedia audio hub opens a route for well-developed, mixed-signal, audio IP to migrate from the hi-fi world into the mobile space. For example, sophisticated 3D surround sound and five-band equalizer functions already established in home cinema and similar A/V and hi-fi systems can be leveraged quickly and at reasonable implementation costs, because much of the design work is already complete.

A more advanced example could be very low noise microphone bias properties, which enable very high quality recording. This would be a unusual feature for a mobile handset, potentially attracting new groups of users. Drawing on high quality audio IP will therefore facilitate unusual and powerful differentiating features with the potential to open new use models and thereby create additional revenue opportunities for the entire mobile supply chain.

Noise performance
Within the tight confines of a modern 3G handset, this high-performance analog audio circuitry must be protected from noise generated by switching regulators, as well as digital noise. An integrated, mixed-signal approach not only relieves circuit designers of this responsibility but also enables a more effective solution. For example, the effects of switching noise can be addressed at the silicon design level by making the power supply switch at those frequencies where the mixed-signal functions are least sensitive to interference.

Tight integration of power management with audio functions also enables the supply to react more quickly, or even anticipate, sudden increases in the system’s power consumption. Such power surges are often caused by volume peaks in audio signals, especially when played through a loudspeaker. Shortening the power supply’s reaction time reduces the need for storing a reserve of electrical charge in large, costly capacitors. As a result, smaller, lighter and cheaper capacitors can be used.

About the author
Robert Hatfield is a technical marketing engineer for Wolfson Microelectronics plc, based in Edinburgh. He can be reached at .