While today the a typical of cell phone is a GSM/GPRS, B/W screen device, or equivalent, the market trends show that devices incorporating color screen, camera phones and personal information management (PIM) applications are growing steadily. For example, in 2006 it is predicted that the number of smart phones, convergent devices with extensive voice and data capabilities, should be bigger than the number of the notebooks shipped that year ""and which far outnumber single function devices like digital still cameras and PDAs.
With this in mind, we have little doubt that newly emerging applications in cellphones and handhelds, such as video streaming and high quality digital media playback, considered today to be exotic, will soon become legitimate in high-end handsets and will later be embraced by the mainstream.
In this article, we will look at the challenges that such complex devices pose, with a special focus on power management. We will also discuss new solutions and future trends.
Growing Complexity and Shrinking Cycle Time
Today's OEMs play in complex markets, spanning across different platforms (2nd generation or 2G platforms such as GSM, TDMA and CDMA, and 3G platforms such as W-CDMA and CDMA2000), and each proposed in different models.
For the best time to market for these applications, the reference design for a single platform typically will rely on a relatively rigid 'core' chipset, while a more flexible periphery will accommodate a model's differentiation within the given platform.
Figure 1. Block diagram of the handset mainboard.
Figure 1 illustrates the core chipset, with the Baseband section, including the application MCU handling the data, the DSP for voice, FLASH memory, the RF section (with its receiving RX and transmitting TX blocks), and the power management unit (PMU) section.
Around such core chipset is a number of add-on modules, such as Bluetooth for un-tethered data transfer on a short distance as in wireless headsets, camera, LCD module and more. These blocks require additional power provided by an auxiliary PMU.
The power management unit
The increasing number and performance of smart loads supported by the PMU demands an increasingly sophisticated power management unit, capable of going well beyond providing the basic functions of voltage regulation, charging and fuel gauging.
In sophisticated systems, the PMU may need to be programmable in order to become platform-specific via software implementation of the protocol, and be capable of communicating with the host CPU via a serial interface (I2CBus or similar). This is to adjust the power delivery mode to the load demand (heavy, light or intermediate modes of operation) and to take responsibility of many critical functions, such as power sequencing including the start-up sequencing, at a time when the communication bus is disabled.
Such PMU can be implemented with varying levels of integration, perhaps initially starting with a solution based on multiple chips for fast time to market, and subsequently up-integrating to a single package (multi-chip package or MCP) or even a single IC, depending on the volumes and other considerations.
Figure 2. Power management unit.
Figure 2 illustrates a microcontroller-based power management architecture providing all the hardware and software functions, as discussed above, in a multi-chip implementation. Many tradeoffs need be considered when defining this unit. The Li+ low voltage (3V typical) power source is conducive to a high level of integration on standard CMOS; but this choice hits a snag if a charger, interfacing with an external AC adapter, needs be integrated, in which case the process technology needs to withstand voltages well above the standard 5V of CMOS.
Some of the regulators, like the buck converter powering the CPU, are required to provide a continuing rising level of power, which may be difficult to accommodate on board of a single CMOS architecture. In this case an external P-Channel DMOS discrete transistor, for example the FDZ299P, housed in an ultra-small BGA package, can help solve the problem. Ultimately, if the cost structure allows for its high mask count, a powerful mixed signal Bipolar-CMOS-DMOS (BCD) process can enable a true single-chip solution capable to handle voltage current and gate count complexity. As illustrated in the figure, each sub-system in the handset requires its own specific "flavor" of power delivery; low noise LDO's in the RF section; and low power LDO's elsewhere. It also requires an efficient buck converter for the power consuming processors; boost converter in combination with LED drivers for the LED arrays; and a linear charger interfacing the Li+ battery with the external AC adapter during charge.
As discussed in the PMU section, the microcontroller, whose block diagram is shown in Fig.3, is the basis of a feature-rich, or smart phone, power management unit. Fairchild's ACE1502 (Arithmetic Controller Unit) family of microcontrollers, for instance, has a fully static CMOS architecture. This low-power, small-sized device is a dedicated programmable monolithic IC for ultraportable applications requiring high performance. At its core is an 8 bit microcontroller, 64 bytes of RAM, 64 bytes of EEPROM and 2/k bytes of code EEPROM, while the on-chip peripherals include a multifunction 16 bit timer, watchdog and programmable under-voltage detection, and reset and clock. Its high level of integration allows this IC to fit in a small SO8 package, but this block can also be up-integrated in a more complex system either on a single die or by co-packaging.
Figure 3. Microcontroller Architecture.
Another important factor to consider when adding intelligence to PMU via microcontrollers is the battery drain in both active and standby modes. An ideal design will provide extremely low standby currents. In fact, the ACE1502 is well suited for this category of applications. In halt mode, the ACE1502 consumes 100 nano-amps, which has negligible impact on reduction of battery life.
As the trend continues toward convergence devices, development of software and firmware becomes an increasingly complex task. In fact, as the systems trend towards larger displays and the inclusion of more functions, such as 3-D games, a phone's processing power and software complexity drive its architecture towards distributed processing. The microcontroller adds further value in off-loading the power management tasks from the main CPU, thus freeing it to perform more compute intensive tasks.
The application of "local intelligence," via a microcontroller, can assume various levels of sophistication, such as the recent trend of "feature phones." For example, it is common to find phones with digital cameras built in to them. But the lack of a photo flash limits the use of the phone's camera in brightly lit scenes. To address this problem, it is now possible to include a flash unit built from LEDs (Light Emitting Diodes). The addition of a flash requires several functions such as red-eye reduction and intensity modulation, depending on ambient lighting and subject distance as well as synchronization with the CCD module for image capture. These additional functions can be easily off-loaded to a peripheral microcontroller. Such an architecture leads to optimized power management, in addition to simplifying the computing load on the main CPU.
Microcontroller-Driven Illumination System
A complex LED-based illumination system is illustrated in Fig. 4. Typically, an array of four white LEDs is needed for the color display backlighting, while another array of four white or blue LEDs implements the keyboard backlighting. White LEDs, typically assembled in a quad package, are needed for the camera flash. And finally, an RGB Display Module provides varying combinations of red, green and blue flashing for "fun" lighting effects. As discussed above, the sequencing and duration of all the illumination profiles are under micro control.
Figure 4. Handset illumination system.
Figure 5 demonstrates the lighting system described above, with all the elements of the system excited at once. (The back light and display light locations are obvious. The flash is the top white light and the RGB is the one in the middle flashing a reddish light.)
Figure 5. Lighting System Demonstration.
The battle for power waste minimization extends to the signal path as well. The logic gates, operational amplifiers and data conversion devices used extensively in ultraportable applications are all specifically designed for ultra low power dissipation and are housed in space efficient packages.
For example, the Ultra Low Power (ULP and ULP-A) TinyLogic devices, such as Fairchild's NC7SP74, a D-flip flop and the NC7SP00 dual NAND gate, operate at voltages between 3.3 V and 0.9V and have propagation delays as short as 2.0 ns. They also consume 30-50% of power compared to existing high performance logic.
Recent high-end handsets exhibit amazing features such as dual color LCD displays, camera, Video on Demand and Audio on Demand. An 800mAh Li+ battery (corresponding to a 2.4Wh at 3V average output) can sustain heavy duty activities like playing games, picture taking and video recording and viewing""assuming each consumes power at a rate of 1.4W for less than 2 hours. Such Figures of Merit (FOM) are getting better, thanks to the power management methods discussed previously, but they remain a far cry from the desired performance of 6-8 hours of un-tethered operation as in more basic handsets.
The two technologies on the horizon promising to improve this situation are Organic LEDs Displays (OLEDs), which do eliminate the power consuming backlights. The other technology is fuel cells, electrochemical devices capable to extract directly electricity from fuels like methanol. Fuel cells already promise to flank Li+, for example as un-tethered chargers, and then to progressively substitute Li+ technology.
Alternative power sources, such as fuel cells, will require even more sophisticated power management. This increased management will necessitate further proliferation of local intelligence to manage tasks (i.e., additional microcontrollers), including sophisticated mixed signal capabilities to perform supervisory functions.
Digital still cameras with OLEDs displays are already commercially available and this technology is expected to take a wider hold in the next three to five years. Fuel cells are a proven technology but difficult to miniaturize and they may come to larger devices like notebooks before trickling down to handsets in volumes. Prototype handsets, some powered and others simply charged by fuel cells, have been demonstrated and are expected to become commercially viable in the same timeframe as OLEDs.
Power management techniques are adapting and evolving to keep up with the increased complexities of today's systems. These techniques include in their cell library traditional regulation elements as well as untraditional digital functions, such as bus interfaces, data converters and microcontrollers.
Feature-rich handsets and smart phones are clearly the devices pushing the edge of every technology, including power, and more features will be coming in the future. For example, it is conceivable that a series of "plug and play" standards will be debated and then adopted to allow for mix-and-match of add-on peripherals (camera, GPS modules, etc.) from various sources, as well as promote re-use of peripherals that a user already owns. The addition of microcontrollers in power management applications will become an increasingly important theme in the ICs that provide system power for these platforms.
This "smartening" of power management electronics, combined with the increasing maturity of new technologies for energy storage and displays, promises to keep these feature-rich devices on a steep growth curve for the foreseeable future.
RENO ROSSETTI is director of corporate strategy for computing and ultraportable for Fairchild Semiconductor International (San Jose, Calif.).
MADHU RAYABHARI is marketing director, Power Management Group for Fairchild Semiconductor International (San Jose, Calif.)<>