Advertisement
Product Releases
Advertisement

Featured Technology Designing Smart Phone Power Management Systems

Thu, 04/01/2004 - 5:49am

LISTED UNDER:

New 3.xG smart phones combine the traditional 2G cellular phone with PDA-like features. Such diverse functionality requires numerous components, most of which have differing power parameters. Therefore, smart, power management is no longer an option — it's a requirement. By Jeff Falin

Cellular phone technology has come a long way since the days of the huge, clunky "bag" phones or those phones that were tethered to the car battery. Today's phones aren't just phones anymore — they're footprint is tiny and they do so much more than just make phone calls.

New 3.xG smart phones combine the traditional 2G cellular phone with PDA-like features, as well as digital still cameras, music players (MP3s) and global positioning systems. Such diversity in functionality requires numerous components, most of which have different power rail voltages, ever increasing current and, therefore, greater power demands. Figure 1 shows an estimate of the increased power demands from a 2G voice call to a 3G video call.

At the same time, consumers want ever smaller phones. This article presents two alternative power management systems that may help smart phone system designers balance the competing goals of increased power requirements from the smallest package, optimal efficiency for maximum battery life and acceptable power rail noise/ripple for the latest generation cell phones.

Choosing the Battery

One of the first tasks when designing a power management system is selecting a rechargeable battery. At present, the only two realistic choices are NiMH and Li-Ion. Li-Ion's volumetric and gravimetric energy densities (270 to 300 Wh/l and 110 to 130 Wh/kg) generally are higher than NiMH's (220 to 300 Wh/l and 75 to 100 Wh/kg). So for the same amount of energy, a Li-Ion battery is smaller and weighs less than a comparable NiMH. In addition, Li-Ion's 3.6-V operating voltage is higher than NiMH's 1.2V.

Most of a cell phone's power is consumed on 1.2 V and 3.3 V rails. To maximize efficiency with switching converters, it is generally more efficient to step down (or buck) from a higher rail to a lower rail than to step up (or boost) from a lower rail to a higher rail. As a result, a Li-Ion battery is the best choice.

Managing the Battery

Proper management and control of a rechargeable battery is critical for maximizing battery life. Battery management consists of three parts: charge control, battery monitoring and battery protection. Charge control ICs have evolved significantly from linear controllers with external pass elements to more efficient switching-based controllers with integrated switches. Battery chargers must handle currents in the 500 mA to 1,500 mA range to provide quick recharge cycles. Battery monitoring and protection ICs are typically packaged with the battery itself. Battery monitoring ICs can be as simple as "coulomb counters," from which the CPU must compute the remaining battery life, to gas gauges with integrated microcontrollers that provide remaining capacity, time to empty, voltage, temperature and average current measurements via a simple communication interface directly to the DSP/CPU.

Power Supply Topologies

The designer must determine the type of power IC, whether an inductor-based switching converter with integrated FETs, inductorless switching converter (or charge pump) or a linear regulator. Each has advantages.

In terms of efficiency, inductor-based switchers have the highest overall efficiency, followed by charge pumps and finally, linear regulators. Cost is generally inversely proportional to efficiency, with linear regulators being the least expensive, then charge pumps and finally, inductor-based switchers. Linear regulators have no output ripple, while charge pumps have some output ripple, and switchers have the highest output ripple of the three. In terms of total solution size, linear regulators are the smallest, typically only requiring an input and output capacitor. In addition to input and output capacitors, charge pumps require one or two additional "flying" capacitors. Switchers require one inductor, which varies in package size.

In 2G phones, there was very little integration of either the digital components, such as the DSP and ADCs, or the analog components such as the power management system. System designers typically chose cost and size over efficiency when designing the power management system. The battery could only be used until its voltage dropped to 3.3 V due to linear regulators only being capable of stepping down their input voltage. Low to medium current linear regulators were used to step down the battery voltage to the remaining power rail voltages in the 3.0 V to 2.8 V range.

In 3.xG phone chipsets, the baseband processor now includes a DSP, a microprocessor/controller, ADCs and DACs for control of the RF as well as the audio signal processing. The processor core voltages are dropping to 1.2 V or below and the I/O and peripheral voltages have begun dropping to the 2.5 V to 3.0 V range. Since the 3.xG phone power rails' current requirements are typically larger than those of a 2.G phone, 3.xG designers need DC/DC converters that provide higher efficiency than a linear regulator, thereby providing longer battery life.

To further prolong battery life, many designers need to use the Li-Ion battery down to its 2.7 V end voltage. In doing so, generating a 3.3-V rail presents a challenge. It seems reasonable that a portable device would have significantly more battery life if the designer takes the battery down to 2.7 V and uses a positive buck-boost or SEPIC converter to provide the 3.3 V rail. But a simplified analysis of a 600 mAh battery (see Table 1) shows that this is not the case. Very little, if any battery life is saved when fully utilizing battery capacity with a SEPIC-type converter instead of stopping the battery at 3.3 V and using a more efficient buck converter.

In addition, when the increased cost of a two-inductor SEPIC converter, or even some of the new, more efficient positive buck-boost converters is considered, stopping the battery at 3.3-V and using a highly efficient switching step-down converter to provide the 3.3 V rail is a valid, possibly even more attractive option. Therefore, the discrete solution presented below uses a buck converter while the integrated solution uses a SEPIC converter to provide the 3.3 V rail.

System Overview

Different components in the smart phone have different power requirements. A simplified block diagram of the primary components demanding power in a cellular phone is shown in Figure 2. For example, the VCO and PLL of the RF section require a power rail with extremely low noise and high power supply rejection to ensure the highest transmit and receive performance. Therefore, although rather inefficient, a linear regulator is the best choice for this rail since it has no output ripple.

It is also important to keep switching frequencies of the DC/DC converters, as well as their 2nd and 3rd harmonics, outside of IF frequency bands. Since DSP/CPU core voltages have fallen to around 1 V, a high-efficiency inductor-based, switching step-down converter is appropriate. White LEDs, used for backlighting the screen, can be powered from either a charge pump or inductor-based step-up/boost converter.

Dynamic Voltage Scaling

Figure 1 shows that the two components requiring the most power are the RF section, primarily the transmitter's PA and the baseband processor. Depending on the phone's proximity to the base station, the PA may consume up to 75% of the total power during a call, compared to only 30% during standby mode. The typical efficiency for older GSM phone transmitters with non-linear PAs is about 50%. But the newer standards like WCDMA require both amplitude and phase modulation that only a linear PA, operating at efficiencies in the 25% to 35 % range, can provide. In addition, the normal baseband processor load requirements for a CDMA2000 1x phone are in the 60 to 120 mA range. So providing the most efficient power to the PA and processor is critical.

Similar to what is used in large-scale integration ICs, dynamic/adaptive voltage scaling DVS/AVS links the processor and regulator in a closed-loop system that dynamically adjusts the digital power supply voltage to the minimum level needed for proper operation. A PA is optimized to have highest efficiency at maximum transmit power. Since most handsets operate relatively close to base stations, the handset radios reduce transmit power to the minimum needed to maintain quality communication. At the reduced power levels, the PA efficiency suffers.

Figure 3 shows that by employing dynamic voltage scaling and adjusting the power amplifier's voltage, efficiency can be increased by 10% to 20%.

Since the power dissipated in a digital processor is directly proportional to the square of the voltage, dynamic voltage scaling can also be used for the CPU. When in standby or some other reduced functionality mode where the CPU can operate at a lower frequency, the voltage can be reduced to lower power dissipation and achieve higher efficiency and longer battery life. For example, consider an OMAP1510 chip powered by a TPS62200 buck converter with a 3.6 V, 1-Ahr Li-Ion battery input and the following characteristics:

DEEP SLEEP (TPS62200 in PFM)

without DVS: Vout = 1.5V @ 300uA,

Efficiency = 93%

DEEP SLEEP (TPS62200 in PFM)

with DVS: Vout = 1.1V @ 250uA

Efficiency = 93%

AWAKE (TPS62200 in PWM): Vout =

1.5V @ 100mA

Efficiency = 96%

Assuming a usage profile of 5% AWAKE and 95% DEEP SLEEP, plotting the output power vs. time shows that by using DVS in DEEP SLEEP mode, battery life is extended by up to nine hours.

The Discrete Solution

A power management system implemented with discrete ICs and with the battery voltage limited to 3.3 V is shown in Figure 4.

In this solution, the buck converter running in 100% duty cycle mode allows the Li-Ion battery voltage to drop almost down to 3.3 V and still provide the 3.3 V I/O rail. Dynamic voltage scaling on the power amplifier and CPU power rail help reduce power consumption by improving each component's efficiency.

The Integrated Solution

The latest process technologies make it much easier to combine, quickly modify, and/or leverage existing discrete IC designs to produce various levels of integrated ICs. For example, generic dual switching converter ICs and dual high-PSRR, low noise linear regulators, application-specific white LED supplies, and cell phone, PDA, and digital still camera multi-rail power management solutions are currently available.

End equipment focused power IC's used in the integrated solution of Figure 5, have integrated peripherals, from ringer and buzzer controls for cell phones to general purpose I/Os for PDAs.

In this solution, the 3.3 V I/O rail is provided by a SEPIC converter, which allows the Li-Ion battery to be drained to its lowest level (approximately 2.7 V). As in the discrete solution, the rails provided by the regulators are taken from the 3.3V rail to improve efficiency. Dynamic voltage scaling on the PA and CPU power rail helps reduce power consumption by improving each component's efficiency.

Discrete vs. Integrated?

In general, an integrated IC is less expensive than multiple, equally rated discrete ICs. Also, an integrated IC uses less board space than discrete ICs performing the same functions. This is due in large part to the extra space needed for trace routing between the discrete ICs. Also integrated ICs may include features, like sequencing of the power rails, vibrator and LED drivers that would otherwise be implemented with discrete IC's.

In the past, integrated ICs have been highly application specific and have not been very flexible. Therefore, they may not accommodate significant design changes late in the design cycle. However, new manufacturing process techniques, including integrated E2PROM for programming output voltage rails and post-package trimming, make "spinning" simple modifications of existing ICs (e.g., an IC with a different fixed output voltage) much simpler, faster and cheaper. On the other hand, the fact that second sources are usually not available for an integrated IC may force the use of a discrete solution.

The Power of Integration

The power solutions presented use power ICs at different levels of integration. The benefits of integrating some or all of the analog power ICs with the digital components, like baseband processor, include even more PCB space savings and reduced overall cost. In the past, one of the hindrances to an even higher level of digital and analog component integration has been the varying requirements of each section of complex electronics. Digital baseband sections require highly dense processes for digital signal processing while the analog baseband and power sections need higher voltage devices. The RF section and specifically the PLL require BiCMOS devices optimized for high frequency operation.

Historically, digital designers oversaw process development and pushed only for high density processes, so circuits requiring high voltage devices were only possible in a different process, meaning a separate digital IC. Recently, semiconductor manufacturers have been developing not only single BiCMOS processes with lower minimum gate lengths for high density and high speed, but also drain extended devices capable of higher voltages for more analog and power applications. Eventually, many of the digital and analog functions, including power management, will be integrated on a single chip.

Future Challenges

Today's consumers are demanding smart phones, with more features and longer operating times before recharging. Newly developed IC fabrication processes have lower leakage currents and lower resistances. This translates into FETs with lower quiescent currents and lower on-resistances, respectively. Ultimately, this results in power ICs with higher efficiencies.

However, unlike constantly-evolving semiconductor technology, battery technology hasn't progressed (and doesn't have anything earthshaking on the horizon) that will allow longer life without increasing the size of the battery.

Of late, advances in capacitor development are beginning to blur the distinction between a rechargeable battery and capacitor. High energy super capacitors are currently being used to power portable devices while the batteries are being swapped. High energy, high power ultra capacitors can provide high current for a short time, and thus provide longer battery life by sparing the battery from having to deliver bursts of energy. These ultra caps are trickle charged during a quiet time and are integrated into the battery pack.

And, there is some talk about fuel cells but, at present ampoules are not standardized.

In addition, fuel cells have poor output transient response. At least initially, fuel cells will be introduced as a supplement rather than a replacement for batteries.

However, improved functionality at lower operating voltages typically imposes more stringent tolerance requirements on the power management blocks as well as low noise layouts. For example, imposing ۭ% tolerance on a 1.2 V power rail requires that the output not vary more than 䕈 mv versus a 3.3 V rail with a ۭ% tolerance being allowed 䖇 mV of variance. Therefore, the demand for DC/DC converters with tighter tolerances, higher currents, greater efficiency and extremely low EMI in as small a package as possible will increase in the next few years.

Conclusion

Various levels of IC integration are simplifying the design of portable power electronics. Specifically, system designers of portable electronics need not worry about managing the power requirements of their devices. Power management ICs at various levels of integration are available to help them maximize battery life, in the smallest board area and at lowest cost.

Jeff Falin is a portable power applications engineer with Texas Instruments. He supports customers who are using LDOs, switchers and charge pumps in applications such as cell phones, PDAs and WLAN cards. He earned a BSEE and MSEE from University of Tennessee. Glossary of Acronyms

ADC - Analog-to-Digital Converter BiCMOS - Bipolar Complementary Metal-Oxide Semiconductor CPU - Central Processing Unit DAC - Digital-to-Analog Converter DSP - Digital Signal Processor DVS/AVC - Dynamic/Adaptive Voltage Scaling E2PROM - Electronically Erasable Programmable Read-Only Memory FET - Field-Effect Transistor GSM - Groupe Special Mobile LED - Light-Emitting Diode IC - Integrated Circuit
Figure 1. Power consumption (Source: Planet Analog) Figure 2. Smart phone power lock diagram. Figure 3. Power amplifier efficiency. Figure 4. A power management system implemented with discrete ICs. Figure 5. Integrated solution.

Topics

Advertisement

Share this Story

X
You may login with either your assigned username or your e-mail address.
The password field is case sensitive.
Loading