Reducing battery discharge rates in CR-2032-based portable embedded designs

Jason Tollefson

September 07, 2010

Jason TollefsonSeptember 07, 2010

The demand for extremely low power technologies was born from necessity, starting with the fabrication process and moving all the way to the application areas. Energy-sensitive applications, in particular, are now driving demand for unprecedented hours of service from a single battery, measured in years rather than days or months.


Meeting this demand requires integrated microcontrollers that have not only been developed in synergy with the characteristics of battery power, but with an understanding of how the application will operate in order to achieve 10, 15 or even 20 years of operation without replacing the primary cell.


This level of extremely lowpower operation demands a new way of looking at applications. Traditionally, the complexity in these applications has been in analyzing the power usage from a convenient viewpoint, making assumptions about the use-case, the application and the power source to generate a "typical" power envelope.


However, to achieve 20 years of service from a single-cell battery, for instance, the application needs to draw less power than the cell's own self-discharge rate, which makes every nanoamp used a critical contributor to the overall power budget (Figure 1 below).

Figure 1: The correlation between MCU sleep current and battery life.

For the industry-standard CR2032, a commonly used lithium/ manganese dioxide cell with a nominal (on load) voltage of 3.OV, the self-discharge current can be as little as 250nA.

In reality, to achieve maximum battery life, the application must use an integrated microcontroller that is capable of operating well under the 1 microampere range during sleep modes, while offering the right mix of processing power, integrated peripherals and on-chip storage at the system level.

When every nanoamp is crucial, it is no longer safe to make assumptions about performance or power. To evaluate the best option, it is necessary to examine parameters that, in non-energy sensitive applications, may not have been critical.

For instance, it is common for extremely lowpower microcontrollers to offer a range of advanced sleep modes, yet it would be wrong to assume that the current drawn in one sleep mode is the same across the entire family of MCUs.

Well known MCUs can exhibit as much as 1,700 percent variation across devices within a family. Therefore, it is important to ensure you start with a family that will offer migration in memory and pin count without sacrificing low-power performance.

Another important aspect is evaluating the device's performance over time, with respect to battery power. As all engineers will appreciate, the way the voltage across a primary cell changes over time is significantly dependent on the cell's construction and the load.

The CR2032, for instance, has a different discharge pattern to a pair of AA/AAA alkaline cells, so an application should be capable of operating equally well under varying battery conditions (Figure 2 below).

Figure 2: An example of varying battery lifetimes.

Without considering the characteristics of batteries, it would be difficult for an engineering team to specify any microcontroller family in an energy-sensitive application and have confidence it will achieve years of functional life from a single battery.

Key items for the team to consider in the design are the consumption characteristics at low voltage, as well as its operational capabilities there.

The MCU should offer operation to 2.OV or below, to extract the most from the battery. Secondly, the MCU should continue to offer operation at higher frequencies at these lower voltages, to maintain the utmost performance of the application.

As the need for extremely low power continues to take hold, the importance of an efficient instruction set architecture (ISA) becomes even more apparent. While energy-sensitive applications may spend in excess of 99 percent of their time in a sleep mode, they will, inevitably, need to wake up at regular or predefined intervals, or in response to an external stimulus.

In this respect, the important attribute is the amount of energy used in order to achieve the task. Design teams now must select MCUs that implement ISAs featuring a greater proportion of single-cycle instructions to achieve a given task, which results in a shorter execution time and a lower energy burn.

This can be illustrated in the following example. If we take the common C function memcpy(), a copy of 32bytes from one memory location to another and compile it for the PIC24F and the MSP430, the resulting code requires 790 percent more clock cycles for the MSP430 (316 vs. 40).

At 3V and 4MHz, this results in 230 percent more energy being expended by the MSP430. The importance of the ISA to power consumption is clearly demonstrated by this example (Figure 3 below).

Figure 3: The number of single-cycle instructions impacts the power consumed.

The embedded electronics industry has reached a watershed event that will change the way integrated devices are now designed, evaluated and implemented in established emerging and future applications.

The significance of this shift cannot be underestimated, and its impact on the industry and industry leaders is already being felt. It is the move to extremely low-power technologies that will become the key enabler for a diverse set of wire-free applications in coming years.

Jason Tollefson is Product Marketing Manager for Microchip Technology's Advanced Microcontroller Architecture Division, specializing in low-power products.


Loading comments...