Handheld devices like cell phones and PDAs rely on a variety of components to perform power management. Maximizing the efficiency of these components, both individually and as a whole, is becoming increasingly important as power consumption climbs (Fig. 1). For example, as cell phones take on new functions like still imaging, video, and Internet access, and move from monochrome to color displays, the drain on the battery rises, decreasing runtimes.
Incremental improvements in the Li-ion batteries used to power handheld equipment have helped to offset some of the increases in power dissipation. But these gains alone aren't enough to maintain the desired runtimes. Consequently, ongoing efforts are made to improve the various power-management components that affect runtime. Battery-management ICs, such as the charger and fuel gauge, are among the components being honed to serve emerging power-hungry applications.
Li-ion battery-management ICs fall into two broad categories. There are those developed for single- and dual-cell applications, which are generally the handheld products like cell phones, PDAs, and digital cameras. The other category involves multicell applications with three or more Li-ion cells. The most familiar example is the laptop, where as many as nine cylindrical Li-ion cells might be found. The requirements for these applications are very different due to the variations in battery capacities, discharge rates, space constraints, and cost.
In general, the laptops require and can justify the added cost of more-efficient charging circuits and more-accurate fuel gauges. However, with handhelds becoming more sophisticated and in need of power, the gap in power-management requirements separating handhelds from multicell applications is closing. Consequently, semiconductor manufacturers are striving to recraft their battery-management components and design solutions. In the process, they're finding ways to migrate some of the battery-management techniques applied to laptops to the handheld devices. In general, they seek to make the new battery-management solutions smaller, more integrated, easier to implement, and less costly. While the focus here is on standalone battery charging components, there also are ongoing efforts to integrate battery charging functions within large power-management ASICs.
Although so many of the battery-management components for handhelds are aimed at high-volume consumer applications, relatively few designers are working on cell phones, PDAs, and the like. Nevertheless, battery-management ICs developed for these products stand to benefit those designers working on a much wider range of products that rely on one or two Li-ion cells (including the Li-polymer types) for power.
With the cost of Li-ion cells tumbling, these high-energy density batteries will take hold in more equipment designs. CD and MP3 players, electric shavers, GPS units, medical monitoring devices, and infusion pumps are among the newer single- and dual-cell Li-ion applications.
In handhelds, battery-charging chips have been developed mainly with small size and low cost in mind rather than efficiency. As a result, the single- and dual-cell battery chargers are more often linears than switching ICs. When first introduced, these linear charge chips required three external discretes—a pass transistor, a reverse blocking diode, and a current-sense resistor. These elements are now integrated within many of the single/dual-cell chargers.
Yet, vendors continue to expand the functionality of these ICs by adding safety, monitoring, and control features. Chip vendors are also developing more compact solutions by introducing new charger ICs in the more space-efficient quad flatpack no leads (QFNs) rather than the larger TSOPs. The QFNs are thermally enhanced with a die attach pad that helps remove heat from the packages, making them suitable for integration of pass transistors.
This feature aids designers in implementing fast charging. Nevertheless, the charger is often embedded in the handheld product, where the heat dissipated by the charger can create problems. One response has been to implement pulse charging, a technique that pushes the power dissipation from the charger's IC back to the ac adapter (usually a wall plug with 5-V output). Although this technique is somewhat controversial (battery vendors don't recommend it), chip vendors claim that it doesn't harm the battery. Another trend in single/dual-cell battery charging is the development of charger ICs crafted to run off the popular USB port.
Meanwhile, many of the same semiconductor vendors are busy developing fuel gauges for the handhelds. These fuel gauges will make it possible to measure battery capacity with much higher accuracy than what most handhelds currently achieve with voltage-based battery monitoring. The improved accuracy will ensure that batteries are charged more fully and permit more accurate detection of low-battery conditions. These improvements will combine to extend battery runtimes in single- and dual-cell applications.
CHARGER IC DEVELOPMENTS
The charge profile for Li-ion batteries consists of a period of constant-current (CC) charging followed by a constant-voltage (CV) charging period. In some cases, when the battery has been overdepleted or when the battery temperature falls outside the cell manufacturer's specified temperature range for fast charging, the CC charging must be preceded by a preconditioning stage. Here, a trickle charge is applied to the battery until the battery voltage or temperature reaches some predetermined set point.
The ability to implement the CC-CV profile is the basic requirement for a single/dual-cell Li-ion battery charger IC. The newer charger chips now offer a variety of other features to help ease charger design, while protecting the cell against potentially dangerous charging conditions. Some popular options are programmable charge current, ac present indication, and charge enable. Charge termination options include safety timers as well as programmable current and voltage limits. Because battery temperature is critical in charging Li-ion cells, some charger ICs also feature a thermistor input to read the internal temperature of the battery pack. Plus, some chargers provide LED or logic outputs to indicate charge status and fault conditions.
The addition of these intelligence and safety features has increased the amount of I/O required. Whereas the first chargers had six to eight pins, the newer ICs tend to have 12 to 16. As with other power-management devices, there's a trend to introduce these parts in QFN packages. For a given footprint, these packages hold a larger die than a TSOP. Even so, another trend is to introduce scaled-down versions of the chargers in packages like the five-pin SOT-23. One of the newer examples of this trend is the LTC4054 from Linear Technology. This charger IC is rated up to 600 mA, which reflects the package's thermal limits. There have also been introductions of charger ICs rated for just a few hundred milliamps. Primarily, these target Bluetooth applications.
That said, much of the charger IC development continues to focus on package types like MSOPs and QFNs. However, even these thermally enhanced packages can't ensure that the charger IC will be able to provide the rated current associated with fast charging during the CC mode.
Consider that when the charger first enters this mode, the voltage across the Li-ion cell could be around 3 V, while the voltage supplied to the charger from the ac adapter is about 5 V. So the linear charger IC sees 2 V across it at a time when it may be asked to deliver a charging current on the order of 1 A. That's asking for 2 W of dissipation in a very small package. This situation may be reflected in the derating of the charger IC on the vendor's datasheet. Some IC vendors implement a thermal foldback technique that automatically senses the charger IC's die temperature and limits the charge current so that the die temperature stays within acceptable limits. An example of a newer device with this feature is Intersil's ISO6291, which comes in a 5- by 5-mm QFN.
Thermal management will remain a consideration as battery capacity rises and the demand for fast charging grows in applications with higher current drain. This could push some applications back to using external pass FETs. The cost of using an external FET will tend to be cheaper because the die for the discrete transistor requires fewer mask steps (eight to 12) than the charge controller (12 to 20). Also, the thermal limitations on the external FET may be less severe.
Higher-capacity cells and higher discharge rates will also affect the system designer's choice of the method of regulation. In handhelds, charger ICs based on linear regulators are generally more popular than those based on the more-efficient switching regulators, because the former types are simpler and cheaper. Nevertheless, some applications that previously used linear charge ICs have migrated to pulse charging ICs to reduce power dissipation in the end product where the charger is embedded.
Rather than applying a steady-state constant current to the battery, the pulse charger applies current pulses with peak levels up to about the 1 C rate. Pulse charging slightly reduces charging time, while keeping the cost of the charger design approximately the same as with a linear charger IC. But pulse charging adds cost to the ac adapter, which then must be current limited. The main benefit of pulse charging is that it effectively pushes the charger's power dissipation back to the adapter, where heat generation is generally less of an issue.
Besides being offered as standalone ICs, pulse chargers can also be integrated within large, power-management ASICs. At least one vendor, On Semiconductor, is now sampling two such ASICs that serve as technology demonstrators. The MCP4100 and MCP4110 each combine a pulse charger with voltage regulators and various portable power-management features.
For some handheld applications, switching charger ICs may offer a better means of reducing the charger's heat generation, despite their added cost. This higher efficiency is exploited in multicell designs, like notebook computers, which have generally required more sophisticated power management than handhelds. But as their power needs rise, handhelds too may warrant use of switching charger ICs.
Still, many single- and dual-cell Li-ion applications continue to rely on linear charger ICs to implement the lowest-cost charger designs. In some cases, this objective is abetted by the input voltage range of the charger IC. These chips are typically designed to operate off a nominal 5-V input. However, some charger ICs have input voltage ranges extending to 10 V or higher.
Although the higher input voltage rating may require a larger, more expensive die for the charger, it allows the charger to operate off of an unregulated or loosely regulated ac adapter. Such adapters cost less than well regulated ones, so the wider input voltage range on the charger IC can actually lower the overall cost of the battery-charging system.
The ac adapter will continue as the main power source for Li-ion battery chargers, but the increasing popularity of the universal serial bus (USB) makes the USB port a viable alternative in some applications. Charger developers are now taking advantage of the USB port's 5-V supply, which provides as much as 100 or 500 mA, depending on the host's capabilities.
In some situations, these USB current limits will hinder the charger's ability to do fast charging. So will the USB's specification of 4.35 V as its minimum supply voltage. Charger ICs developed with USB in mind must take into account the variability in the current and power available from USB sources. These chips will also contend with other USB specifications, such as those for quiescent current, reverse current, and inrush current. While a variety of Li-ion charger ICs offer the option of charging off the USB port, not all of these chips are necessarily compliant with the latest USB 2.0 specifications.
TRACKING BATTERY CAPACITY
In most handheld devices, the battery-management circuits implement a rough form of fuel gauging based on voltage monitoring. This approach adds no additional components to the system and doesn't increase power dissipation. Voltage monitoring is commonly used to detect low-battery conditions. Yet this technique is very inaccurate because the battery voltage corresponding to a selected level of capacity varies with battery discharge rate, temperature, and aging.
The difficulty of picking a voltage point at which to issue a low-battery warning is illustrated by a graph of battery voltage versus discharge rate and temperature (Fig. 2). In this graph, the voltage corresponding to a fixed amount of remaining battery capacity varies widely over the discharge rate and temperature ranges shown. Consequently, when a fixed-voltage point is selected for low-battery warning, the actual capacity for that voltage point will vary a great deal over temperature and discharge rate (Fig. 3).
To account for the variations in capacity for a fixed battery voltage, large capacity guardbands are required when voltage monitoring is used to issue a low-battery warning. Unfortunately, that means low-battery warnings will be issued too soon or too late in some cases.
With the advent of cell phones offering color displays and Internet access, power dissipation has risen dramatically and reduced battery runtime significantly. In addition, the consequences of inaccurate fuel gauging are higher now that data can be lost when the battery runs out. As a result, there's greater demand for more-accurate fuel gauging in handhelds.
To obtain this higher accuracy, designers will need to adopt coulomb counting. When combined with an accurate fuel-gauging algorithm, coulomb counting permits measurement of battery capacity to within 1% of actual capacity. That's because this technique can account for capacity variations due to changes in discharge current levels and temperature. Although fuel gauge ICs have been in use for multicell Li-ion applications such as notebook PCs, these devices have not met the cost and size constraints of the handheld equipment.
However, semiconductor vendors have responded to the new requirements for capacity measurement by developing fuel-gauge ICs, also referred to as gas-gauge ICs, aimed at the single-cell applications. (Most handhelds use just a single Li-ion cell, not two.) These fuel-gauge ICs will compete with each other on the basis of their accuracy, ease of implementation, power dissipation, cost, and package size. Size is particularly important because the fuel gauge and its associated components often must fit in the battery pack along with the cell's protection circuit.
One recently introduced single-cell Li-ion battery gas gauge IC is Texas Instruments' bqJunior, also known as the bq2700x. Packaged in an eight-lead TSSOP, this fully integrated gas gauge measures the battery's charge and discharge currents (via an external sense resistor), battery voltage, and temperature using an on-chip analog-to-digital converter (ADC) and voltage-to-frequency converter. Dynamically balanced, the chip can measure and correct for its own offset. Like other fuel gauges, it requires a calibration step when installed in the battery pack.
The bqJunior's measurements are fed to an on-chip processor, which executes a coulomb-counting algorithm to calculate the remaining battery capacity and system runtime. The host simply reads the gas gauge's data through an HDQ interface. That data includes available power, average current, temperature, voltage, and time to empty and full charge. This device will be priced at $2.55 in quantities of 1000.
Other vendors have also indicated their plans to introduce fuel-gauge ICs for single-cell applications in the near future. One such company is Intersil. Unlike TI's approach, the Intersil fuel gauge will rely on the host microprocessor for capacity and runtime calculations. The company will employ an ADC to sample battery data, which will then be passed over an I2C bus to the host. According to Intersil, this approach can reduce power consumption by an order of magnitude versus having a microcontroller on-board, while also permitting use of a smaller die and package.
Nevertheless, another vendor plans to develop an on-chip-microcontroller-based solution that will integrate a fuel gauge with the battery protection circuit and the battery tag. (The latter function identifies the type of battery in the pack.) This device, which is slated for launch at the end of the second quarter, will take advantage of a very low-power microcontroller that consumes less than 100 nA in standby.
1. "Power Management Challenges in Hand-held Products: A Holistic Perspective" by Ed Bordeaux, Staff Applications Engineer, Intersil; Power2002 presentation.
2. "New Handheld Devices Need Accurate Battery Management" by J. Norman Allen, Vice President Strategy, Microchip Technology; Power2002 presentation.
|Need More Information?|
International Components Corp.
Peter Sadler, (312) 850-8621
Mark Richey, (760) 803-0477
Linear Technology Corp.
Dave Bell, [email protected]
Maxim Integrated Products
Naveed Majid, (408) 530-6044
Jim Vernon, (469) 241-0087
George Paparizzos, (650) 215-5255