The Digital Handset
10 February 2009
In mobile phone architectures, the evolution of speaker drive technology has been struggling to progress, due to the levels of integration in audio system-on-chip devices.
Combined with the limitation of interchip audio signal transmission to the analogue domain, this can limit
flexibility and innovation in the final stages of audio data processing.
While consumers are looking for more power output and better quality from their mobile phone speakers, the lack of access to digitised audio data streams off-chip has made it difficult to use the latest digital audio
enhancement techniques.
Digital audio enhancement algorithms for voice and for music can bring many benefits to the user and can be broadly grouped into three categories; improving sound quality for voice and music, making speakers sound louder, and extending battery life during music playback or voice calls.
Attempting to apply audio enhancements of this kind whilst having to work around modern architectural limitations tends to have some unpleasant side effects. These are increased power consumption and
shorter battery life, increased silicon and passive component footprint on the PCB, degraded audio quality, and reduced flexibility for experimenting and optimising audio algorithms.
However, there are signs that these audio architectures are now starting to change, and greater access to digital audio data offchip presents an opportunity for applying powerful digital enhancement techniques
which can improve speaker performance for music and for voice.
Integration constraining innovation
Many highly integrated processors and audio SOCs integrate audio DACs in an attempt to add value and save space, though in practice they can often have the opposite effect,
adding cost, increasing passive component count and limiting audio communication within the handset to the analogue domain.
The analogue interconnect between the SOC and the speaker driver causes a range of common problems such as increased passive component count, gain errors which limit output power, and the possibility of noise pickup in the long analogue connections.
There is usually an additional cost of analogue switches or mixers which are required for routing cellular voice to the same speaker driver and for suppressing pop noise. Also, consider how an audio electronics
designer might attempt to apply speaker enhancement functions to the signals from both the multimedia
processor and from the baseband processor using this architecture.
Firstly, the signal needs to be re-digitised to allow algorithms such as speaker equalisation or 3D stereo enhancement to take place. Whilst crude analogue implementations of some of these features are currently available in some speaker drivers, they simply cannot compete in the same league as digital implementations when it comes to flexibility, cost-effectiveness and performance.
If these digital speaker enhancement algorithms were to be implemented using today’s architectures they would look similar to figure 2.
However, there are several problems with this arrangement. In most cases the processor or SOC will not
use leading edge DAC technology as product developments of this kind often focuses on digital feature innovation, with mixed signal feature addition as an afterthought. This means that the finer points of audio DAC design are often overlooked. High power consumption, poor noise performance, pops and clicks and poor output driver efficiency are common limitations of this kind of device. Once noise performance has been degraded it cannot be recovered later. Similarly, power dissipated in the processor’s DAC can not be
regained later. The re-digitisation in the speaker driver burns yet more power in the ADC, which needs to be of high performance to avoid degrading SNR or THD any further, and is expensive in terms of power and
silicon area. The need for accurate clocking of the ADC places further overheads on this implementation.
Digital connections
As baseband processors are now beginning to use digital data transmission for voice within the handset, there is an opportunity for multimedia processors and SOCs to move in the same direction and use digital
connections to the devices that drive loudspeakers.
A more effective architecture is shown in figure 3. This allows a much greater degree of freedom to innovate without the higher cost, larger footprint, higher power consumption and signal degradation suffered by existing architectures. Two types of audio enhancement in particular can now be made available to the designer, both of which effectively address the end user demands for higher quality, louder speakers and longer battery life. These are dynamic range control and parametric equalisation.
Dynamic range control brings many benefits. It can provide basic safeguards against speaker damage and thermal overload of the speaker driver. Secondly, by limiting the maximum power output, the battery is protected from excessive current draw, which helps prevent clipping during periods of battery droop and so reduces distortion during large signal transients. By reducing peak current in this way, the battery is also protected against premature shutdown, so playback time is extended. This reduced distortion and longer battery life bring clear benefits for the user but these can be improved even further if the dynamic range control can be tailored to the acoustics of the handset. This flexibility is enabled by using digital transmission to the audio DAC or CODEC which drives the speaker.
Another benefit of digital dynamic range control is the ability to make the small speakers that are used in mobile phones ‘punch above their weight’ in loudness. By increasing gain during quiet periods, the
intelligibility of movie soundtracks, which were never designed for playback on a mobile phone, can be improved. The apparent loudness of a ringtone can also be improved.
Improving voice call quality
A further advantage of this architecture is the ability to improve voice call quality. Most dual-processor multimedia phone designs are effectively dual-mode gadgets, capable of operating as either portable media players or as a basic phone, with shared access to the speakers and headset via analogue
switching or mixing. This means that even the limited digital audio enhancement algorithms, which might be available on the multimedia processor, are not available at all during a voice call.
Due to the increasing proliferation of digital audio transmission between baseband processors, multimedia processors, and audio CODECs in mobile phones, there is an opportunity to take a leap forward in speaker enhancement technology.
By locating speaker-specific digital enhancement features such as equalisation and dynamic range control in the device which actually drives the speaker, and by using digital audio interfacing between devices, most of the limitations of today’s architectures can be bypassed effectively.
In 2009, Wolfson will be releasing a number of products that will provide designers with the tools to create these new architectures. The benefits of these are clear for any end user wanting louder speakers, longer battery life and improved audio quality for voice as well as music.
ROBERT HATFIELD is Principal Product Architect, Wolfson Microelectronics
Contact Details and Archive...
Most Viewed Articles...