Selecting the best battery for embedded-system applications

As systems and their corresponding power-management strategies become more sophisticated, so, too, does the task of selecting a battery that suits both static and dynamic requirements. Appropriate battery selection involves chemistry, temperature, peak demand, energy transfer, and environmental concerns.

Keith Odland, Silicon Laboratories Inc -- EDN, November 18, 2010

Another type of primary battery, zinc-carbon, is a predecessor of and similar in composition to alkaline technology. This low-performance battery addresses cost-sensitive applications that do not require high performance, such as toys, alarm clocks, and radios. These batteries are readily available to OEMs. The nominal voltage of these cells is 1.5V, with a discharge voltage of 0.9V.

A third type of battery is lithium, which most commonly uses the designation BR or CR. BR lithium batteries come in a variety of form factors but are most commonly available as coin-cell batteries. Manufacturers typically fabricate them using a carbon monofluoride gel and a lithium alloy. This composition has good high-temperature characteristics, and the batteries usually have low self-discharge characteristics. As such, they find use in applications that require long service intervals and have relatively low power requirements. Such applications include water and gas meters, heat-cost allocators, electronic toll-collection systems, and tire-pressure-monitoring systems. These batteries are also readily available to OEMs. The cell’s nominal voltage is 3V, and its discharge voltage is 2.2V.

Like BR lithium batteries, the CR type uses a lithium alloy for the anode but replaces the cathode with a manganese-dioxide material. This material reduces the internal impedance of the battery. As such, a CR cell generally better suits supplying higher pulse currents than its BR counterpart at the expense of a slightly higher self-discharge rate and lower performance at high temperatures. Applications include remote keyless entry, RFID (radio-frequency identification), and watches. Both OEMs and consumers can easily obtain these batteries. Their nominal voltage is 3V, and discharge voltage is 2.2V.

Lithium-thionyl-chloride batteries are relatively new and have extremely low self-discharge rates, enabling battery life of approximately 20 years. They also benefit from a flat discharge profile over time so that the terminal voltage stays relatively constant over the entire service life. Manufacturers of these batteries typically fabricate them using a solution of lithium tetrachloroaluminate in thionyl chloride as the liquid cathode, with a zinc alloy as the anode. This technology is more costly than other lithium chemistries and finds use in applications demanding extremely long battery life, such as water and gas meters and other industrial- and military-electronic applications. These batteries are uncommon in consumer applications and are available to OEMs through a select set of suppliers. The cells have nominal voltages of 3.6V and discharge voltages of 2.2V.

Zinc-air batteries provide a much higher energy density than the previously noted types. They receive their power by oxidizing zinc with oxygen from the air with the help of a hydroxide-based solution. Consumers are most familiar with this type of battery for hearing aids and camera batteries; however, much larger versions of this type find use in marine- and railroad-navigation applications. These batteries have a shelf life of multiple years. Once they are in service, however, they last on the order of hundreds of hours in consumer applications. Both OEMs and consumers can easily obtain them. These batteries have nominal voltages of 1.4V and discharge voltages of 0.9V.

Battery suitability

Engineers assess a number of parameters when evaluating the suitability of a battery type for an application. Some of the most common factors they use in this analysis include nominal voltage, energy capacity, energy density, self-discharge capabilities, and dynamic considerations. The nominal voltage is the voltage as measured across the battery’s positive and negative terminals. Engineers often partition multiple batteries in series or in parallel to provide a more desirable cell voltage or current supply to the application.

Energy capacity is the stored energetic content of the battery. The SI (Système International d’unités) unit for energy is joules, but most battery manufacturers specify it in milliamp hours. Because the total energy in a battery is a function of both the amount of current that the battery can source and the terminal voltage, using joules is a more consistent way of comparing batteries with different chemistries. You can easily convert a battery capacity from milliamp hours to joules with the following equation: E=C×VT×3.6, where E is the energy in joules, C is the capacity in milliamp hours, and V_T is the terminal voltage.

Battery chemistries rely on electrochemical reactions to provide electrical energy. Some of these reactions are more potent than others, which can lead to the development of small batteries with the same energy content as some larger counterparts. This size-to-energy ratio is the energy density. As a general rule, the higher the energy density, the more costly the battery technology is. Designers constantly struggle to find the optimum balance of cost and energy density.

Batteries do not last forever. Even if they sit unused on a shelf, electrochemical reactions are still taking place, slowly diminishing the batteries’ energy content. This naturally occurring process is the self-discharge rate. Alkaline batteries generally have a service life of seven to 10 years. Lithium BR and CR batteries have a service life of 10 to 15 years, and lithium-thionyl-chloride cells can last more than 20 years. Self-discharge rates and other deteriorative mechanisms affecting battery life can highly depend on temperature and duty-cycle characteristics. Fluctuating duty-cycle requirements often can have an adverse effect on the ultimate discharge characteristic of a battery.

Dynamic physical parameters also affect battery performance. Variations in temperature, output impedance, duty cycle, and energy delivery affect battery-loading conditions and ultimately shape the battery-selection process. Some of these variations are first-order effects, and you must give them appropriate consideration.

Many systems have high dynamic bandwidth with respect to power demand. For example, a wireless sensing system in an advanced-metering-infrastructure-class gas or water meter can have dormant power consumption on the order of microwatts and an active peak consumption of watts. In other words, dynamic-system-power-demand bandwidth can be microwatts during low-duty-cycle sleep mode and watts during active and high-duty-cycle radio-transmit mode. This situation creates additional power-delivery requirements that the battery must accommodate alone or with another energy-storage device. To solve this problem, engineers often place a capacitor in parallel with the battery to provide peak energy demand. In these cases, they must also consider the additional design issues of the capacitor’s cost, size, charging scheme, and leakage.

In addition to the noted considerations, engineers must pay attention to battery-discharge profiles, which can vary greatly depending on the battery chemistry and the power-demand profile both peak load and duty cycle. Environmental considerations, especially temperature, can affect battery performance. System-level considerations, such as battery-replacement interval and system-voltage requirements, also influence battery selection. Environmental considerations, such as recycling, toxic materials, heavy metals, safety, and shipping regulations, are of concern, as well.

Meeting requirements

As with most other engineering problems, designers must weigh a set of sometimes-conflicting requirements to develop the best approach to meeting system specifications. To illustrate this point, consider an exaggerated example. Law-enforcement personnel sometimes use stun guns, which are high-voltage, nonlethal electronic-control weapons, to incapacitate combative subjects who pose a risk to a police officer, an innocent citizen, or themselves. These devices deliver thousands of volts to the assailant’s body, temporarily disrupting the nervous system and rendering the individual unconscious. The system uses a transformer, among other techniques, to step up the battery voltage thousands of times higher than its original terminal voltage.

Instead of using a transformer, a device designer could choose to achieve the design objective by arranging 30,000 AAA alkaline batteries in series. This design would also be able to deliver a 45,000V shock to the assailant, along with other obvious practical limitations of using an approximately 0.8-mile-long (1.33-km-long), 792-lb (360-kg) stun gun—not to mention the 50-kV thumb switch!

Although an exaggeration, this example highlights the fact that, by using modern electronics, you can overcome some of the natural limitations of the battery’s electrical chemistry and use them in different ways. As another example, zinc-air batteries have long found use in hearing aids because of their energy density of 1.69 MJ/kg and ability to deliver high peak currents. The batteries typically have a service life of less than three months due to the time it takes the electrolytic reaction to reach its conclusion. However, this service lifetime is acceptable for the application, and the batteries come in “calendar” packs so that the user has a new replacement battery for each month.

Another aspect of this battery chemistry is that the terminal voltage of a single cell is typically 1.4V. Specialized low-voltage circuits for hearing aids address this limitation, but the battery voltage does not translate easily into mainstream embedded-system electronics. You would need to make additional provisions to make a nominal 1.4V cell useful for standard CMOS embedded electronics.

Fortunately, more devices are integrating advanced power-management units to address these challenges. For example, a chip with an integrated dc/dc boost converter could boost the 1.4V input voltage of a zinc-air battery or the 1.5V input of a common alkaline battery to an appropriate value for the system.

More important, a dynamically programmable boost converter could change the output voltage depending on the needs of the system so that the energy from the battery to the system always operates efficiently. This operation would enable system engineers to optimize the power-supply efficiency for the use case during runtime.


For example, a bidirectional wireless sensor node for a home-security application could be a glass-breakage sensor with a bidirectional communication link comprising a transmitter and a receiver. This sensor monitors the condition of a window and periodically reports the status of the window and the battery to the main control panel. The communication between the sensor and the control panel uses a transmitter/receiver/acknowledgment protocol, which reduces the number of redundant messages the sensor sends to the panel. Most of the time, the sensor is in a low-power mode to maximize battery life. Table 2 defines the states the glass-breakage sensor can occupy.


If you look carefully at the system assumptions, it becomes clear that dynamically adjusting the battery voltage optimizes power efficiency and performance. For example, you can obtain maximum transmitting-power efficiency when the transceiver is operating at 3V. The alkaline battery has only a 1.5V nominal terminal voltage, so you can achieve 3V operation with the integrated dc/dc boost converter, yielding approximately 90% efficiency. However, internal regulation limits the receiver chain to 1.8V. Supplying 3V during the receiver transaction would cause the internal low-dropout regulator to reduce efficiency to 60%. It would be better to dynamically adjust the output of the dc/dc converter from 3 to 1.8V and increase the efficiency during the sensor’s receiving transaction.

Compare the system using a lithium coin-cell battery with a fixed-voltage rail with an alkaline battery using the dynamic-switching technique. The switching loss is 0% using the coin cell because the switch-mode supply is not in use and the terminal voltage is 3V. You also may not need to increase the size of the coin cell to meet the peak current demand because this resizing would require the use of a large, expensive coin cell. In the alkaline-battery case, you can assume a 10% switching loss.

Tables 3 and 4 detail the energy each element of the wireless-sensor application requires. The sleep duration is 1 second minus the sum of all of the other transactions. The processing, receiving, and transmitting functions occur once per minute. Table 3 shows the requirements for a 3V, 620-mAhr-rated, approximately 62-cent CR2450 lithium battery, and Table 4 shows the requirements for a 1.5V, 1125-mAhr-rated, approximately 25-cent AAA alkaline battery.

With this usage profile, the CR2450 battery would last approximately 4.33 years. With the same usage profile, the AAA alkaline battery would last approximately 4.65 years. This duration represents 16% higher efficiency, resulting in a 7% increase in service life with a 60% decrease in battery cost. This comparison demonstrates the gains you can make by using more modern dynamic techniques of energy conversion. These gains highly depend on the duty cycle of the functions that derive the greatest benefit from the high-efficiency power supply. As the receiver-mode duration or duty cycle increases, so do the benefits of using the alkaline-battery approach with the switched-mode supply. The dc/dc converter’s output could be 3.3V—that is, 0.3V higher than that of the lithium battery—and could provide greater output power and enhanced range.

Given the growing sophistication of battery technologies and chip-level power-management techniques, the industry has come a long way since Alessandro Volta discovered the voltaic pile in 1800 (Reference 1). More than 200 years’ worth of technological evolution and innovations in the fields of chemistry, electrical engineering, and manufacturing have resulted in batteries thousands of times more sophisticated in design and function than when Volta invented the original versions. Today’s system designers now have many more options when selecting the appropriate battery to support their next embedded-system designs.

---

Reference

“Voltaic pile,” Wikipedia.