Optimizing Power Management in Multimedia Mobile Devices

By: By Mark Jacob, Director of Marketing for Audio and Power Management, Dialog Semiconductor
( 1 May 2009 )

The earliest mobile computers in the 1980s were introduced without any recognizable power management. The device was simply either on or off with board power supplies provided by rudimentary zener diode-referenced voltage regulators. System size didn’t matter much, nor did wasted heat. Over time the regulators became more precise by integrating more control transistors for enhanced loop feedback. Eventually switching topologies were utilized to bring about significant power dissipation reductions, especially at increased peak output currents. The use of discrete off-the-shelf power management components provided the necessary flexibility to early products.

NOW IT’S COMPLICATED
Contrast these approaches with the system power management circuits of today. In some lower cost devices, basic power management has been integrated into the baseband processor, but high-end multimedia, mobile devices now employ dedicated power management ICs (PMICs) comprising over one million transistors, unique resistors and capacitors to provide system power management. It’s no longer simply a question of providing a regulated power supply rail of a given voltage and current capacity. Your mobile phone might have in excess of 30 distinct power domains, all to be derived from a single battery.

More granular power domains help minimize battery drain and extend operating life between recharges for the end-user. This practice, however, creates more complex power sequencing – the order and timing of power supplies being switched on and off – and you need to be able to set up multiple sleep modes so that parts of the system not in use at a particular time don’t consume power. Power control has to be adaptive, reacting rapidly to changes in the power requirements of each system function. This can often be achieved using combinations of linear and switching topologies. In fact, power management has become sophisticated to the point that it can be optimized depending on the nature of the signals passing through parts of a system. Added to this, the same power management circuit can be responsible for monitoring, protecting, and charging the system’s main battery.

HALF THE SIZE OF POWER MANAGEMENT FUNCTIONS
PMICs have now replaced discrete MOSFETs, DC/DC converters and low dropout regulators (LDOs) in more advanced handheld, multimedia products. Their obvious benefits, versus using discrete power management devices, are reduced board space by up to 50 percent in some instances, smaller bill-of-materials, and reduced design complexity. These are often the main drivers for adopting PMICs. However, there are a number of other advantages that can be equally important:
1. System reliability is improved by minimizing the number of external components and connections, and through better thermal management. The use of thermal vias within the PMIC ensures effective heat transfer to ground connections and thermal runaway is prevented through integral thermal shutdown protection.
2. High-density, mixed-signal designs present board layout challenges. Layout-related design problems cannot all be identified with simulation tools and a significant proportion of overall product design time is consumed in fixing cross-talk, decoupling, current and voltage sensing, and EMI problems. You reduce these problems greatly with an integrated power management device, therefore reducing time-to-market.
3. Integrating power MOSFETs and regulators onto a single piece of silicon means that higher switching frequencies can be used. This is because the MOSFETs are closer to the regulators, so lead inductance and gate-drive impedance are reduced. These higher switching frequencies mean that smaller filter components – inductors and capacitors – can be used on the board, saving even more space.

PROCESS SIMILARITIES PROVIDE OPPORTUNITIES FOR AUDIO INTEGRATION
Clearly, power management is a mixed signal function. Voltage and current levels are analog in nature but power path and system control demand digital precision and flexibility.
In broad terms, many of the same silicon fabrication processes and process geometries that apply to power management functions are equally applicable to mixed-signal audio processing. This makes it possible to take the benefits of power management integration one stage further with the integration of audio CODECs onto the same chip. This approach to system partitioning also facilitates optimizing the power consumption of CODECs depending on the different audio streams that the system uses. The DAC in the PMIC takes the decoded voice, uncompressed audio or ringtone data from an external processor, and provides conversion, amplification and output to earphones or a speaker. To optimize efficiency, the voltage may be scaled up or down depending on the required bit rate and signal-to-noise (SNR) performance demanded. The same PMIC may integrate microphone channels using low noise automatic gain control amplifiers and ADCs. At all times, unused blocks are switched off to conserve power.

As in almost every aspect of system design, there is a trade-off. The close proximity of audio and high-speed digital signals compromises the audio signal-to-noise ratio with a typical 10-15dB SNR degradation. However, this is more evident when integrating audio functions into baseband chipsets. Baseband process constraints limit the success of this approach to system partitioning, and they will become more of a problem as process geometries for digital circuits continue to shrink.

A typical PMIC of the kind described above is shown in Figure 2. Designed to complement digital baseband cellular processers, it integrates a 24-bit HiFi stereo DAC, a voice CODEC with 64-256Kbps sampling, 15 LDOs, three buck converters, two boost converters, LED and vibrator drivers, a touchscreen interface, and two audio drivers for the speaker and headset. These functions are controlled by a host processor through an industry-standard I2C interface.

A further advantage of process-based partitioning – in other words, putting multiple mixed-signal functions onto a single device – is that developments in the digital parts of the system platform can be implement without a complete re-design. Faster and more feature-rich applications processors and baseband processors can be more easily adopted, with reduced development and test times, and fewer type-approval hurdles.

In addition, the PMICs described are programmable, so they can be used across complete families of end product, thereby supporting a platform approach to mobile system design.
It is now accepted practice that power management needs to be based on a holistic, system level approach. How mobile system designs are partitioned has a significant impact on the effectiveness of power management, and on system size, cost and complexity. With battery life and system functionality being key determinants of consumer choice in mobile devices, it’s clear that optimizing power management is a critical consideration in creating a winning consumer product.

WHAT’S COMING NEXT?
PMICs that offer even greater integration and flexibility are now beginning to appear that are significantly more complex than those with integrated audio CODECs. These platform-PMICs add configurability to programmability, enabling a single device to support multiple applications and mobile graphics processors. They completely manage the energy flow into and out of the battery and all major board components. With increased autonomous regulation, on-chip LDOs may be cascaded to DC/DC converters to further improve system efficiency. What’s more, you can connect outputs in series or parallel, adding further flexibility. These devices represent today’s most advanced power management within a single (7mm x 7mm) IC and will be available in volume later in 2009.

Click here for the illustrations:

Figure1, Figure2