It's time to Appcessorise!

Ultra low power (ULP) wireless is on the up. Proprietary 2.4GHz technologies have steadily carved a growing niche in applications as diverse as wireless desktops, remote controls and fitness monitoring sensors such as heart rate belts. But it is the introduction of interoperable technologies, allowing OEMs a choice of chips from a multi-vendor environment that are guaranteed to work together, that’s really going to push this technology into the mainstream.

Technologies such as ANT, from Canadian developer Dynastream Innovations, and Bluetooth low energy (a part of the latest Bluetooth v4.0 specification) are receiving support from major consumer electronics manufacturers such as Apple, Microsoft and Sony-Ericsson. Devices from these manufacturers incorporate software and hardware that make it easy for developers to design ultra low power wireless peripherals that connect seamlessly with the companies’ PCs, tablets and smartphones.

These peripherals plus an associated downloadable app are fuelling a new market for so-called ‘appcessories’. An example is a blood glucose monitor incorporating Bluetooth low energy paired with a Microsoft Surface tablet computer. The diabetic’s blood glucose readings could be automatically transferred from the meter to the tablet via Bluetooth v4.0 and an app could indicate trends and perhaps highlight the risk of a hypoglycaemic (low blood glucose) episode before it occurs. Better yet, on their next visit to the specialist, the diabetic’s monitor could transfer all its data to the physician’s Windows 8 PC, allowing him to make a rapid and accurate assessment of the condition. But this application represents just a drop in a huge ocean of potential.

Analysts have a bit of a reputation for exaggerating the potential of new technologies, but even if they are only partially correct about the potential of the appcessories market it will still be huge. According to consultants Juniper Research, for example, the market for wireless accessories linking to smartphones and tablets will reach 110 million units a year by 2017.

Longer Life

The key to ULP RF technology’s success is that it extends wireless connectivity to devices that — for cost, size, weight or accessibility reasons — are unable to support the batteries needed for technologies such as classic Bluetooth (versions prior to v4.0) or Wi-Fi. Even ‘low power’ wireless technologies like ZigBee and low-power Wi-Fi consume too much energy for typical ULP applications.

A typical power source for ULP wireless applications is a 3V, 220mAhr CR2032 coin cell battery. While these devices feature impressive energy density, there is not a huge power budget to play with. And yet, consumers will not tolerate changing the battery in their heart rate belt or wireless mouse frequently. The design needs to cater for months of service or even up to a year.

Assuming a device such as a heart rate belt is used for an hour a day, the average current draw on the battery will need to be 220mAhr/365 x 1hr = 600µA for it to last a year. But there is much more to it than that; the pattern of demand influences how much of that initial 220mA will actually be available to the application and it’s likely to be quite a bit less than the data sheet would have you believe.

ULP wireless applications are characterised by short periods of activity when the radio wakes up to transmit or receive data followed by much longer periods of inactivity. Duty cycles of just a few percent are typical. Of course, it depends on the application, but a 1ms burst of activity at a frequency of 40Hz is a good example. So the loading on the battery is a pulse pattern with a peak current draw dictated by the radio, microprocessor and any other component that needs to be powered while communication is underway. Unfortunately, drawing power from a coin cell in this way, particularly if the peak current is high, is a sure fire way to limit its capacity; even if the average current is in the microampere range that might otherwise be assumed to give good battery life.

Maximising Accessible Capacity

As is the case with all batteries, a lithium coin battery’s available capacity (measured in milliamp hours (mAh)) is dependent on the rate of discharge. Figure 1 shows the voltage level of a CR2032 coin cell battery as it is drained with continuous currents ranging from 500µA to 3.0mA. When discharging the battery at a rate close to the rate stated in battery data sheets (< 500µA), it can be seen that discharge follows the familiar curve and that the full 230-to-240mAh capacity can be accessed before the voltage drops too low. However, as the rate of discharge increases, the effective capacity in the battery drops.


Figure 1: CR2032 coin cell continuous discharge battery voltage curves

The accessible capacity — referred to as the Functional End Point (FEP) — will also be dependent on the voltage level. This is the minimum voltage level the electronic circuitry can operate at and is therefore a parameter you decide when choosing active components for your application. Let’s assume that your application has an FEP of 2.0V. As seen in Figure 1, at 500µA a CR2032 can deliver its full capacity of 240mAh before reaching this FEP. But if the battery is drained at 2.5mA, the FEP is reached after only 175mAh of the battery capacity is used.

So much for continuous currents; ULP wireless application are characterised by pulse loads as the radio switches on to receive or transmit and then goes back to sleep. The pulses vary depending on the application and radio technology but for the purposes of illustration let’s consider a fairly typical pulse period of 25ms (1ms ‘on’/24ms ‘off’).

Figure 2 shows the drain curves when subjecting the battery to 10, 30, 50, and 80mA peak currents under this 25ms pulse cycle. At 80mA it can be seen that both the high peak current and the high average current challenge the battery. The 2.0V Functional End Point (FEP) is reached after approximately 70mAh. As the peak current drops, the 2.0V FEP creeps closer to 200mAh, but even at 10mA the peak drain is still too high for the battery to be able to deliver the full 240mAh capacity specified in the data sheet.

Figure 2 shows that, as long as the application can be designed with a FEP of 2V or lower, and peak currents lower than 30mA, the battery capacity loss is ‘manageable’, providing around 175 to 185mAh. But if the FEP is higher (> 2.0V), the impact from the higher peak current rapidly increases. For example, if the application has a FEP = 2.4V the battery can still deliver 175mAh when the peak current is restricted to around 10mA, but the capacity drops to 100mAh if the peak current is around 30mA.


Figure 2: Peak current magnitude impact on battery capacity for CR2032 with a 25ms pulse cycle

Work Fast And Go Back To Sleep

Sometimes the answer to conserving energy can be counter-intuitive. Selecting a more powerful processor, for example, would seem to be a way to burn more power. But it depends on what you do with that processor.

Nordic Semiconductor earned its reputation designing ULP 2.4GHz transceivers. When the company designed its new range of wireless connectivity solutions, the nRF51 Series, the engineers wanted to improve on the competitive power consumption performance of the previous generation products but take advantage of the computational power of a 32-bit microprocessor instead of the 8-bit device used in the previous generation.

The processor core chosen for the nRF51 Series was the 16MHz, 32-bit ARM Cortex-M0 processor. Although the ARM core consumes about 4.4mA when executing code from Flash — a similar amount to the enhanced 8-bit 8051 core found in the previous generation chips — its average power consumption is lower than the older processor because a fast start-up (up to 100 times quicker at just 2.5µs) and rapid execution of complex code (the ARM core has 10 times the processing power of the 8-bit device) allows the ARM core to return to a low power sleep state very quickly. And, as we’ve shown, average power consumption — along with peak currents — is a key factor in determining battery life.

Apart from using an ARM core, the nRF51 Series included some other tricks to keep the system power down. For example, Nordic implemented a bus technology called Programmable Peripheral Interconnect (PPI) that allows the processor to initiate a peripheral function yet return to a low power sleep state while the peripheral devices complete the operation autonomously.

Further small power savings come from the new 2.4GHz radio that offers a 9.5dB improvement in radio link budget while consuming less than 10mA RX/TX peak current (using the on-chip DC-to-DC converter with a 3V supply). That compares with the 13mA peak current of the previous series.

Three power supply options are available: an on-chip DC-to-DC buck converter operating from a 2.1 to 3.6V supply; an LDO for 1.8 to 3.6V operation, and an LDO-bypass mode for operation between 1.75 to 1.95V, maximising the available capacity of the CR2032 coin cell. The end result is a chip that exhibits up to a 50 percent lower average current consumption than the company’s previous generation of products (see Figure 3).


Figure 3: Power savings for the nRF51 Series compared with Nordic’s nRF24L Series. Faster processor core, more efficient radio, and power-optimised architecture add up significant average current savings

Designing For Ultra Low Power

If a designer is limited to drawing power from a modest coin cell, the design must work in the most efficient manner possible and so requires careful component choices. If the design is based round a 2.4GHz radio chip, it is essential to check the data sheet for both peak and average currents in applications similar to the target use case.

The typical pulse load of a ULP wireless application plays havoc with the battery life if the power peaks push up to 20 or 30mA. Also bear in mind that radio chips with a raw data rate of 250kbit/s have to be on air for much longer than devices with megabit bandwidths to transmit equivalent payloads, which pushes up the average current.

Beyond the radio chip consider maximising battery life by increasing the FEP margin by selecting active devices (ICs) that have as low a minimum supply voltage as possible, minimising the current drain by selecting devices with the lowest peak current, and designing the power management to prevent multiple high drain devices (for example RF chip, LEDs and LCDs) loading the battery at the same time.

Then identify the worst-case drain in the application by first ascertaining in which mode of operation the activity is highest and therefore the battery demand greatest. Search the IC data sheets for power consumption during normal operation and periods of RF activity.

Next, estimate your corrected battery capacity (which is likely to differ significantly from the nominal capacity in the data sheet); use this corrected battery capacity and the application’s highest average current consumption to estimate the worst case battery lifetime using the formula:



Note that as long as your application can stay in a mode with a lower average current consumption than that used in the calculation, it will be able to operate longer, but depending on how much battery capacity is left, it may fail once it again enters a mode with higher average current consumption.

Finally, remember that these rules apply for any high peak but pulsed battery load, not only wireless/RF circuitry. The high peak pulsed loads could also be from any combination of LEDs, vibrating motors, piezoelectric buzzers, LCD backlights and so on.

Author profile: Jay Tyzzer is a Senior Field Application Engineer with specialist 2.4GHz ultra-low power transceiver maker Nordic Semiconductor. He is responsible for engineering support in the Americas. Tyzzer has also worked with OEMs such as Space Vector, Pertec and Tandon Magnetics and tier one distributors, Hall-Mark, Hamilton Avnet, and Symmetry Electronics. He received his engineering degree from Piece College and is a member of IEEE.

Photo courtesy of: Wahoo Fitness