Part of the  

Chip Design Magazine

  Network

About  |  Contact

Headlines

Headlines

Mobile Device Power-Management Strategies

Ultra-low-power MCU technologies enable today’s always-available mobile devices.

Consumers’ love affair with always-available mobile devices shows no signs of waning. But as reliance on these devices grows, innovative power management can be a significant product differentiator. We talked to Evan Schulz, applications engineer for microcontroller products at Silicon Labs and Pradhyum Ramkumar, product marketing manager for MSP430 ultra-low power micro-controllers at Texas Instruments to get their advice on a range of new approaches.

EECatalog: Mobile device users demand long battery life along with instant-on response. What are some strategies developers can use to balance those conflicting requirements?

schulz_evan

Evan Schulz, Silicon Labs: To provide an instant-on response, developers need to use a truly interrupt-based approach. An interrupt-based application requires a wide range of wake-up-sources as well as a fast wake time. MCUs with a wide array of wake sources allow the MCU to stay in sleep mode as long as possible. The longer the MCU stays in sleep mode, the lower the power consumption and the longer the battery life. In addition, developers must use a device with a fast wakeup time. When a wake source triggers a device wake, the fast wake time provides the instant-on response that users demand. Compromising flexible wake sources or fast wake time results in decreased battery life or responsiveness.

ramkumar_pradhyum

Pradhyum Ramkumar, Texas Instruments: Consider a pedometer or indoor navigation application on your phone. The phone measures the number of steps you walk each day or uses sensors to determine your location inside a building. Another example is a battery charging solution in which the processor needs to probe the battery charger, fuel gauge and make decisions on when to charge the battery and when to switch power sources. In these applications, the mobile phone will need to continuously gather data even while it is in your pocket or purse. The application processor in these cases is continuously active, counting the number of steps and monitoring sensors, gauges and, in essence, wasting battery life.

While applications processors implement different power modes, the fundamental issue is the need to power up core modules even to complete simple tasks that require little or no processing—like reading a sensor or charging a battery, etc. With complex operating systems like Android running on the applications processor, the switch between power states not only involves waking up the processor, but also increased latency associated with re-initializing the operating system. Also, most application processors have smaller geometries, and hence higher leakage current, which affects standby power.

Using an “always on” ultra-low-power microcontroller as a co-processor to the application processor solves this problem. The application processor can be completely shut down and the co-processor can wake up periodically to read sensors, gather data, pre-process it and only wake up the main processor when necessary. The co-processor not only has lower standby current (due to larger geometries), but also has faster wake-up times.


EECatalog: With new silicon technologies that enable low-power states, how are developers using software to get even more performance out of the system?

Schulz, Silicon Labs: MCUs with a variety of low-power states can provide many optimization “knobs” for developers. Sophisticated power-aware software tools from MCU suppliers allow developers to utilize complex power consumption models and highlight configurable parameters to fully optimize their MCUs for the lowest power. In addition, power models can help developers identify trade-offs between low-power settings. For example, a developer can identify the power savings associated with running the system 10 MHz slower without ever measuring the current or reading the datasheet. Both of these features can help the embedded designer reduce development time and allow the MCU to be fully optimized for ultra-low power.

Ramkumar, Texas Instruments: First, wake up the processor only when needed. Sleep most of the time, wake up quickly, process data and go back to sleep.

Second, use intelligent peripherals that gather data while allowing the processor to sleep. For instance, a serial port with DMA functionality can work while the processor is asleep. Once adequate data is buffered, an interrupt can wake up the processor to process the data.


EECatalog: As developers look at reducing bill-of material counts through greater integration, how is that affecting power management strategies?

Schulz, Silicon Labs: In some respects, higher levels of single-chip integration are making power management easier. When more functionality is integrated into the MCU, developers are able to isolate higher-current areas and optimize as needed. Chasing down excess current draw in an embedded system can be a very difficult task. As external components are moved into the MCU, developers have fewer potential areas to investigate. For some functions there can be synergy through various peripherals that developers can take advantage of. For example, analog components such as oscillators, comparators and analog-to-digital converters (ADCs) can have built-in modes that are optimized for the low power modes of the MCU and automatically handle the complexities around entering and exiting the various power modes. In summary, integrating more peripheral functions into an MCU can help ease the developer’s efforts to manage power for the lowest possible current.

Ramkumar, Texas Instruments: While more sophisticated process technologies offer smaller geometries and a higher level of integration, they also increase the amount of leakage. As a result, while active power has reduced, standby power has actually increased. In case of applications in which the processor is asleep most of the time, this may actually be a disadvantage.

Also, going to very small geometries on mixed signal processors (processors with analog and digital modules) is quite challenging as analog does not scale in size as digital. Hence mixed signal processors tend to not be in the smallest process nodes, unlike applications processors.


EECatalog: Mobile devices are increasingly becoming “hubs” for a wide range of accessories—personal fitness monitors, home or industrial automation appliances, or diagnostic tools from medical to automotive. What are some of the power challenges faced by the accessory designer, as well as the smartphone or tablet designer to enable this?

Schulz, Silicon Labs: Increasingly, consumer, medical and industrial devices need a means for communicating with each other. Hub designers require a communication protocol that the accessory designers can also use. Depending on the communication medium, power consumption can become a significant issue. A wireless link between two devices will consume much more power than a wired link, but a wired link might not be a viable option for most mobile devices. If the accessory needs to be charged by the hub, the hub’s power budget must encompass charging. The accessory developer needs to be mindful of the time required to charge the device as well as the expected lifetime on a full charge. It is important for the hub to be designed with expansion in mind, or the accessories become limited and will outgrow the device. In most cases, it is less expensive to upgrade the accessory hardware than it is to upgrade the hub hardware.

Ramkumar, Texas Instruments: As previously mentioned, the mobile device constantly needs to “listen” to all the signals around it to be able to make intelligent use of the data and respond. For example, phones today can wake up as soon as they are picked up (using sensors) or gripped, or possibly when a voice prompt is provided. A “listening” processor obviously consumes more power than one that is not. Handling these stimuli is possibly better handled by an ultra-low-power co-processor rather than the main applications processors to conserve battery.

Likewise for an accessory, how does it get to know that a mobile device is in sight without continuously looking for it? What if the accessory was not charged and loses power just when you need it?


EECatalog: As the cloud continues to take over the world, requiring always-connected, constantly updated applications, how will developers adapt to this constant power drain?

Schulz, Silicon Labs: Developers have three primary strategies to handle always-connected devices in a cloud computing environment:

  • •Optimize the design for lower power
  • •Use larger batteries
  • •Provide more charging stations/options

I believe that the most likely adopted strategy is a combination of lower power optimizations and larger batteries. As technology evolves, the power consumption of portable cloud-connected devices will continue to decrease. In addition, as the cloud continues to grow, non-volatile storage will not be as important in the design of the device. Designers can replace the physical space occupied by non-volatile storage with larger batteries. Larger batteries combined with more sophisticated power optimization will enable longer operating times.

Ramkumar, Texas Instruments: First, not all data needs to be transferred instantaneously to the cloud. For instance, in a pedometer application, it does not make sense to update a cloud service with every step. Second, just like in video/audio transmission, instead of sending raw data, mobile devices will send processed/compressed data to conserve battery power.


coupe_cheryl

Cheryl Berglund Coupé is editor of EECatalog.com. Her articles have appeared in EE Times, Electronic Business, Microsoft Embedded Review and Windows Developer’s Journal and she has developed presentations for the Embedded Systems Conference and ICSPAT. She has held a variety of production, technical marketing and writing positions within technology companies and agencies in the Northwest.

Tags: , , ,

Leave a Reply