Putting Designs on a Power Diet

Between battery replacement and recharging cycles, I sometimes feel like I’m constantly nursing power-exhausted personal electronics back to full function. Even though I keep a reasonably watchful eye on power status, it’s not unusual for my fitness wearable or Bluetooth earbuds to die during a workout, and I won’t even reprise a familiar refrain about smartphone charge levels hitting empty at the worst possible moment.

Extrapolate an individual’s experience with a handful of personal electronics to an Internet of Things (IoT) application with thousands of battery-powered devices, and it’s easy to imagine those applications collapsing under the weight of battery maintenance alone. For those large-scale IoT networks and personal devices alike, the desire for more immediate data from “always-on” sensors magnify the power problem. Fortunately, this bleak image of power-starved electronics is becoming less likely as silicon manufacturers hone the power efficiency of microcontrollers and shift some of the processing burden away from the main processor.

Classic power management augmented by advanced technologies

The conventional view of microcontroller-based systems focuses largely on the duty cycle of the main processor because it is typically responsible for most of the power consumption in small embedded systems. Designers have been taught to work to minimize the amount of time the processor spends in its highest power-consuming active state. Instead, a power-constrained system is designed to allow the processor to rest in its power-conserving sleep mode as much as possible. For applications requiring periodic data collection from sensors, developers let the processor sleep and use peripheral interrupts to wake it just long enough to collect and process the data before returning to its sleep state.

The emergence of more sophisticated on-chip peripherals has let developers extend the time the processor spends in sleep state. Microcontrollers routinely integrate peripherals like analog-to-digital converters (ADCs) that are able collect sensor data without waking the main processor at all. Semiconductor manufacturers have further extended this concept in more advanced microcontroller architectures built to support intermediate power modes between fully active and fully asleep. In these devices, the intermediate power modes can selectively enable the various separate power domains for the processor core, on-chip memory, analog peripherals, and digital peripherals.

Advanced processor families such as Maxim Integrated’s Darwin microcontrollers take this approach to the next level with an extensive set of mechanisms designed specifically to reduce power consumption without compromising application functionality and performance requirements (see “Build More Effective Smart Devices: Part 1 – Low-Power Design with MCUs and PMICs”). As a result, developers can more finely balance power and performance to meet tight power budgets.

Peripherals get their own processors

In separating peripheral functionality from core processing, more advanced microcontrollers have enhanced those peripheral subsystems with dedicated processors. For example, Maxim Integrated’s Darwin series, like many devices in this class, includes a peripheral management unit (PMU) that goes beyond the usual support for direct memory access (DMA) operations with the inclusion of round-robin scheduling and other more advanced functional capabilities.

This diffusion of processing capability beyond the processor core forms the foundation of some of the most effective power reduction and performance enhancing approaches available today. One obvious example of this trend lies in the cryptographic hardware accelerators built into most advanced microcontrollers designed for the IoT or other connected applications. By speeding algorithm execution, dedicated accelerators allow the device to return more quickly to a low-power state.

A more interesting example of this trend arises in wireless microcontrollers such as the Texas Instruments SimpleLink family. For example, the Texas Instruments CC2640R2F Bluetooth low energy (BLE) wireless microcontroller combines an Arm® Cortex®-M3 main processor with a dedicated BLE subsystem comprising a dedicated Arm Cortex-M0 processor and radio frequency (RF) transceiver (Figure 1).

Figure 1: Advanced wireless microcontrollers such as the Texas Instruments CC2640R2F BLE device optimize power consumption by using an energy efficient Arm Cortex-M0 processor core to maintain wireless connectivity while the Arm Cortex-M3 main processor sleeps. (Image source: Texas Instruments)

While the main processor runs the application, the Cortex-M0 processor is unavailable to the developer and runs only the BLE protocol stack. Because the power-efficient Cortex-M0 core can continue to operate at low power levels while the main processor is in sleep mode, this microcontroller can provide always-on connectivity without overburdening tight power budgets.

The need for always-on functionality is of course not just a requirement for connectivity. In a growing number of sensing applications, users expect their devices to respond instantly to changes in temperature, movement, air quality, and other characteristics. With conventional methods, this always-on functionality would effectively force the microcontroller to operate in active mode continuously, or nearly so, while collecting and examining data for significant events.

Many advanced sensors let developers program minimum and maximum thresholds for triggering an interrupt, allowing the microcontroller to remain in sleep mode until the threshold crossing event occurs. In some applications, however, even this threshold capability isn’t enough. An always-on motion sensor, for example, may need to recognize characteristic changes or patterns in measured acceleration or orientation that indicate the device user is walking, running, climbing stairs, turning, or other activities. Even with advanced threshold-capable sensors, the host microcontroller would need to remain active to identify these characteristic changes. Instead, the STMicroelectronics LSM6DSOX sensor module is capable of identifying patterns of interest thanks to its built-in finite state machine and decision tree processing engine.

For developers, capabilities like autonomous peripheral operations, dedicated processing engines, and local sensor processing are only a few of the methods available to help put battery-powered designs on a power diet.

Reference:

Build More Effective Smart Devices: Part 1 – Low-Power Design with MCUs and PMICs - https://www.digikey.com/en/articles/techzone/2018/oct/build-more-effective-smart-devices-part-1-low-power-design-mcus-pmics

About this author

Image of Stephen Evanczuk

Stephen Evanczuk has more than 20 years of experience writing for and about the electronics industry on a wide range of topics including hardware, software, systems, and applications including the IoT. He received his Ph.D. in neuroscience on neuronal networks and worked in the aerospace industry on massively distributed secure systems and algorithm acceleration methods. Currently, when he's not writing articles on technology and engineering, he's working on applications of deep learning to recognition and recommendation systems.

More posts by Stephen Evanczuk