Consumer demand for flexible audio features and hi-fi sound from mobile phones has led to handset audio chip designers having to optimise the Class D amplifier to meet the noise and power management demands. Robert Hatfield explains.
Designers of mobile products, such as multimedia mobile handsets, are today faced with demands to provide versatile, high-quality audio functions, including high-output speakerphone modes and simultaneous speakerphone/headphone operation. These are necessary to deliver a wide range of music and video-oriented services to hungry subscribers.
To make this possible, more is required of the audio amplifiers in mobile multimedia devices. But since the efficiency of the Class AB amplifiers traditionally found in mobile handsets does not exceed 20-25 per cent in most practical situations, a small increase in output power comes at the cost of a large increase in current consumption. Today's mobile consumers will not accept shorter battery life, a larger and heavier battery - and certainly not both.
High-efficiency mobile amplification
The Class D amplifier offers an answer to this challenge by achieving much higher efficiency than Class AB amplifiers.
Operating on switching principles, rather than exploiting the linear portion of the output transistor characteristic, a Class D amplifier is vastly more efficient than a Class AB amplifier of similar output power.
In addition, the latest Class D designs are now able to achieve very low total harmonic distortion (THD) comparable to high quality Class AB performance.
However, implementing a Class D amplifier suitable for mobile applications is no trivial challenge.
In addition to the known issues, including output filter design and management of switching noise, the size constraints prevailing in mobile introduce a new set of issues associated with integrating a high frequency switching amplifier alongside sensitive analogue audio, and other mixed-signal and analogue functions, on the same chip.
Class D operation, in a clamshell
Class A, Class B and Class AB amplifiers are linear amplifiers, in which the output transistors are operated in their active regions.
In the case of Class A, the transistors in the output stage are operated continuously in their active regions, leading to relatively high current consumption in return for excellent linearity. The maximum efficiency, which occurs only at the maximum signal level, is no better than around 25 per cent.
In a Class B amplifier, high-side and low-side transistors operate alternately in a push-pull configuration. The theoretical maximum efficiency is much higher than Class A, but is much lower in practice. However, non-linear operation in the crossover region introduces significant distortion.
The Class AB amplifier uses the Class B topology, but the transistors are biased to always operate in the linear region. This eliminates the Class B crossover distortion, at the cost of slightly increased power consumption.
In contrast to these linear amplifiers, the Class D type is a switching amplifier.
The audio signal to be amplified is input to a comparator, which compares it with a sawtooth wave. The result is a pulse width modulated (PWM) square wave where the period is equal to that of the sawtooth, while the pulse width represents a sample of the audio signal. The sawtooth frequency is set very much higher than the maximum audio signal frequency.
The PWM square wave - and its inverse - then drive a Mosfet H-bridge, turning opposite Mosfets on and off to set up an alternating current that represents the square-wave sample of the audio signal.
Because the output transistors are either turned hard on or off in order to steer the current, the only losses in a Class D system are Rds(on) losses in the Mosfets, and other resistive losses, hence the system is very efficient.
Integration challenges
The small footprint of modern mobile devices places extra constraints on audio design, including the implementation of Class D amplifiers.
The squeeze on dimensions demands highly integrated audio design, which has already led to mixed-signal integration of codec, power management and speaker output functions. This level of integration brings significant challenges in managing noise performance, for example.
When a Class D amplifier is introduced to the mix, the noise management challenge moves into another gear. Switching noise, for example, although attenuated in the Class D output filter, must also be prevented from corrupting the audio signal path in the analogue portion of the chip.
Power supply decoupling, which is important in any Class D design if switching noise is not to degrade circuit operation, takes on a greater importance in the context of mixed signal chip design.
Emerging systems
When implementing a multimedia codec featuring integrated Class D output stages, assiduous noise attenuation techniques are required. Particular attention is paid to providing a stable and clean analogue rail, usually between 2.7V and 3.0V.
Further careful attention to noise management also demands measures to minimise the effects of GSM signalling noise on the audio output. In GSM handsets, the RF power amplifier is powered up at 217Hz intervals, to send data to network basestations. This makes the audio path susceptible to audible buzzing and clicking at the 217Hz switching frequency. Very high supply rejection at 217Hz results in significantly improved audio performance.
External components and wasted power should also be minimised, to meet the demands of mobile applications.
For instance, modern high output speakers are usually connected directly to the battery to maximise output power. This has additional implications when integrating codec, amplifier and speaker driver functions.
Firstly, since the DAC is operated at a substantially lower voltage than the speaker, the signal must be level shifted to drive the speaker sufficiently hard. A gain stage that relies on external components adds to the bill of materials, board layout complexity and footprint, leading to a preference for level shifting entirely on-chip.
Secondly, direct battery connection increases the potential for leakage capable of quickly running down the battery. Low leakage design will extend standby time.
Finally, some system designers may see problems with switching noise during certain modes of operation, for example, when the handset is being operated as an FM radio receiver. An accumulation of system design parameters may conspire to produce excessive interference in the receiver. Dynamically selectable Class D or Class AB amplification modes give system designers extra flexibility to optimise audio performance under all likely usage scenarios.
Demand for features dominates the mobile design landscape, but energy management has the ultimate power of veto. This fact has been the major influence in drawing Class D amplification into the mobile arena.
Designers are under pressure to deliver greater multimedia capability for fewer Watts. It is effective, but optimal implementation demands good audio and mixed-signal design skills, to create an integrated system that will meet small footprint, low component count and low leakage demands implicit in any mobile system requirement.
Robert Hatfield is technical marketing engineer at Wolfson Microelectronics
See also: Electronics Weekly's Focus on Mobile Linux, a roundup of content related to the open source operating system shaped for mobile devices.