Though there is heightened interest in wearables, in its simplest form, the technology can claim a history of more than 45 years. The first examples appeared in the early 1970s – digital watches from companies such as Casio and Pulsar. Such products were very different from the smartwatches that are being brought out today. While they offered much less functionality, battery life was much greater. And this is one of the real issues with modern wearables – as new generations of products have evolved with increased feature sets, the operational run times that can be expected have shortened dramatically. To put it bluntly, things are going in the wrong direction.
The rudimentary wearables of the past connected directly to the battery, which ran the microcontroller and display; there were no additional layers of power management to be concerned about. The battery would keep going for three years or more, which meant they were convenient for users and there were no unexpected shutdowns. With the latest wearables, the big challenge is that, in many cases, battery life is less than a day. As a result users cannot be sure that their wearable device will remain operational long enough for them to get the opportunity to recharge it. This is not only leading to user frustration, but also holding back the market to some degree.
There are specific characteristics inherent to wearable device design which have major implications for its battery reserves and design engineers need to be aware of these. Firstly, available space for batteries will be limited. Secondly, with the consumer electronics business being highly competitive, bill of materials costs need to be kept as low as possible. Thirdly, though the demands being placed on batteries has increased considerably, the fundamental chemistry has not. So ever more sophisticated power management is needed, with efforts to raise power conversion efficiency levels and to reduce power losses.
Wearable technology can take various forms, so let’s focus on a common example – a fitness tracker. This will consist of:
- A microcontroller for processing tasks
- A display to give the user access to data
- A low power wireless transceiver
- Various sensors, and
- A supporting power management system.
Each represents a drain on the battery and how big this drain is will be critical to the tracker’s operation. Technological advances mean each item is gradually drawing less power. A TFT draws less current than previous display technologies and, as we move to new process technologies, MCU power requirements are being scaled down. In most cases, the wireless transceiver will be compliant with the Bluetooth Low Energy (BLE) protocol. Each item will only be operational for a small proportion of the time – for example, the display is only fully active when the user is abstracting information from it, while the BLE transceiver is only needed during data transfer. The MCU is going to be functional for a reasonably large proportion of the time, however, and power management circuitry is going to be in use all the time. Though it seems counterintuitive, conventional power management mechanisms can be the biggest drain on power during a wearable device’s working day.
As computational capacity has got cheaper over the years, the temptation has been to keep adding more functionality onto the MCU. But, when it comes to power sensitive applications, such as wearables, it is clearly not the right thing to do because turning on the MCU and other power hungry components as little as possible is critical to the power budget. In some cases, MCUs targeted at wearable designs can draw really low standby currents, which makes them look appealing. But they can take 4 or 5 mA to wake them up and if they have to be woken up on a fairly regular basis, this is going to drain a sizeable chunk of the battery.
A new strategy needs to be implemented. The combination of an ultra-low power MCU with a programmable power management unit allows ‘distributed intelligence’ to be made use of and improvements in overall power efficiency to be derived. It means that only the functionality required is drawing power from the battery, extending battery life. In our example of a conventional wearable fitness tracker – a low power MCU, BLE transceiver and sophisticated power management chip –power consumption is only 3 or 4µA when it is inactive. However, when it has to wake up, to either check for motion or sniff for RF, power consumption maybe 200 times that figure.
If a programmable power management unit (PPMU) is incorporated and given the capacity to take care of these wake/sleep cycles via its internal timers, instead of the power management chip/MCU, then the power budget can be ramped down, as the MCU does not need to get involved. The PPMU can also take responsibility for collecting data from the sensors, so the MCU is woken up only when required for calculation purposes. The PPMU can even carry out basic human machine interface functions and, as a result of following this approach, it is possible to reduce the overall power profile of wearable devices by up to 30%.
Implementing a distributed intelligence model means the entire system can run in a more power efficient manner than simply integrate more functionality onto a single chip. In the same way as has been seen in building automation, this topology is the way forward. By migrating some functional elements from the MCU and letting the power management element deal with them, it is not only possible to use a smaller battery, but also to extend the battery’s life span. Moore’s Law has allowed the semiconductor industry to keep pushing to ever smaller process nodes. This has resulted in increasingly integrated ICs, which have brought many benefits. However, this has not come without a degree of compromise – and that has been most noticeable in the power consumed. The long term progression of wearable technology will depend heavily on further advances in power management. It seems clear that implementation of distributed intelligence will have a big part to play, by moving away from the conventional thinking that has kept pushing towards higher and higher degrees of integration.
Engineers need to make better architectural choices, rather than doing what they have always done and perpetually looking to add more and more onto the chip as greater provision became available. We must stop being so lazy about how we go about things and start applying a power-led perspective, rather than a purely integration-led approach.
Author profile:
Matthew Tyler is director of strategic business development for ON Semiconductor.
