Internet of Things (IoT) sensor nodes often need to be able to sustain short bursts of activity. However, the sudden changes in demand from the battery can deplete the cell faster than expected, causing problems for remote installations, and may demand the use of a larger, more expensive cell than is necessary. The supercapacitor provides a way to deliver the charge needed to support brief periods of burst activity that does not stress the battery subsystem. This article examines the role of supercapacitors in IoT designs and explains how their chemistry allows them to provide support for burst activity.
The IoT promises to greatly increase the intelligence of systems that deal with the environment, transport and industrial control through the use of highly distributed semi-autonomous sensor and actuator nodes placed precisely where they are needed. One of the key elements in an IoT sensor node design is its energy autonomy: using energy harvesting, long-life batteries or a combination of both to ensure a lifetime of five, ten or even twenty years. For this class of design, every millijoule counts, demanding careful attention to detail.
Over the lifetime of a sensor node, the average power consumption will be extremely low, which makes it possible in some cases to provide all the energy required from a harvester, such as a photovoltaic (PV) module. However, the average power consumption hides the impact of the sensor node’s duty cycle.
Typically, the IoT node will spend much of its time in a low-power sleep mode, waking only periodically to take a sensor reading. Very often, microcontrollers employ hardware assistance to reduce the number of times that the processor core itself needs to be started to execute software. A dedicated state machine for the sensor interface can be set up to take the reading and then only send an interrupt to the MCU if it is outside a programmed limit. If that occurs, the MCU can wake and perform further processing and if, in turn, it is an important change that needs to be reported, a wireless message can be sent out to a remote server via a low-power radio-frequency connection.
The resulting power-consumption profile resembles a series of pulses of varying height depending on the number of modules that need to be active at any one point. Although the current consumption is to a large extent predictable, it does exhibit large swings over the lifecycle of the system. Battery cells and energy harvesters do not necessarily work at optimum efficiency in this type of environment.
The sudden changes in demand from a battery can deplete the cell faster than expected, causing problems for remote installations, and may demand the use of a larger, more expensive cell than is necessary. For example, lithium thionyl chloride primary cells are often used in wireless sensor nodes because they have high capacities per gram and exhibit low self-discharge – which can be less than 1 percent per year – which contributes to a long service life. Discharging those batteries with higher current can result in lower energy efficiency than expected, mostly because of the losses from internal resistance.
Supercapacitors, otherwise known as double-layer capacitors, provide a good match for designs that have ‘bursty’ power demands. Their ability to reduce the peak power demands on the chemical battery are also likely to extend its lifetime, a critical factor in consumer electronics devices, although there is a clear tradeoff between the energy lost through self-discharge versus the problems caused by continual battery charging.
A supercapacitor employs a double-layer construction in which two non-reactive porous carbon electrodes are immersed in an organic electrolyte. When a voltage is applied to the plates, the potential on the positively charged plate attracts negatively charged ions, and positively charged ions move to the opposite plate. The movement results in two separate layers of capacitive storage, one at each plate.
The use of a porous carbon-based electrode material with a structure that gives it a surface area on the order of 1000 m² per gram provides very high plate areas, and therefore capacitance. Supercapacitors also benefit from having a very small nanometer distance between the plate and the stored charge, which is controlled by the size of the ions in the electrolyte.
The combination of large surface area and very small charge separation gives the supercapacitor its increased capacitance compared to purely electrostatic components. However, there is no actual chemical reaction: the electrochemical changes are purely based around the polarization of the electrolyte. Because the changes are purely based on polarization, the process is highly reversible, enabling supercapacitors to be charged and discharged hundreds of thousands of times.
Supercapacitors also have low equivalent series resistances, which makes them able to give up their energy very quickly to create very large currents. Supercapacitors are also less likely to be affected by temperature than chemical batteries, since they’re not reliant on a chemical reaction that could be slowed by low temperatures. The parts have been used in applications that need to be charged in seconds and then discharged over a period that may last for seconds up to many minutes. They are often used as backups for a power supply to ride out intermittent drops.
In an IoT application, the supercapacitor can provide a buffer between the battery and the rest of the circuitry, supporting the short-term demands of the MCU and wireless subsystem without putting too much stress on the battery itself. Similarly, the supercapacitor can act as a charge reservoir for systems that rely on environmental harvesting, building up sufficient energy over a period of time to sustain short-term bursty needs. Alternatively, the supercapacitor can be used to prevent excessive power demands from the DC/DC converters used to condition power from the battery or harvester when the system is quiescent but needs to sustain some functions during the long periods of sleep.
As an energy reservoir, the supercapacitor has several key advantages over the battery. A supercapacitor can be charged and discharged an almost unlimited number of times in contrast to the electrochemical battery, which has a defined cycle life. There are some ageing effects. Under normal conditions, a supercapacitor fades from the original 100 percent capacity to around 80 percent over ten years but this is less of degradation than with an equivalent secondary battery under normal conditions. As well as the ability to support many charge-discharge cycles, the supercapacitor’s lower internal resistance allows it to deliver high currents up to the point of full discharge.
There are drawbacks to the supercapacitor, primarily its rate of self-discharge which is much higher than that of a chemical battery. The stored energy of a supercapacitor decreases from 100 to 50 percent over a period of about a month. A nickel-based battery, which is not designed for longevity, self-discharges by 10 to 15 percent per month. Self-discharge is the main reason why this type of device is not yet viable as a primary energy source for IoT sensor nodes.
In a typical IoT application, the self-discharge problem will not be encountered directly, as the component will be supporting cycle times that are on the order of seconds or minutes. However, the leakage will need to be factored into predictions of energy usage and battery lifetime.
There are two ways in which the supercapacitor can be harnessed to smooth out power demand from the primary sources. One is as a reservoir for peak current demands. In this configuration, where the capacitor is placed in parallel with the battery in order to ride out current peaks, the harvester or primary cell will top up the capacitor after bursts of activity.¹
An alternative or an addition to the use described above is to allow more of the sensor-node electronics to shut down during periods of sleep. Most designs will use some form of switching the DC/DC converter to regulate the voltage to the MCU and sensor circuitry. Although these components can provide very high operating efficiencies, particularly in comparison to low dropout (LDO) regulators, their performance is often not optimized for the ultra-low-load conditions encountered in sleeping sensor nodes.
Figure 1: Typical usage of a supercapacitor in circuits where the supercapacitor is used to power the sensor node while the DC/DC converter is inactive.
The current demands may be on the order of the power wasted by the DC/DC converter itself even in low-activity modes such as pulse skipping. This can make the decision over whether to also switch off the DC/DC converter during sleep an important consideration. In this case, a supercapacitor can be used to provide the trickle of current required by the real-time clock and related circuitry.²
The supercapacitor's leakage circuit will play a major role in the decision over the topology to employ as the leakage could approach that of an efficient DD/DC converter, particularly if the downstream circuitry requires some level of regulation to overcome the problem of the supercapacitor's output voltage dropping as it discharges. Leakage for supercapacitors is normally quoted in terms of the current needed to maintain a voltage after a period of approximately 30 minutes.
There is now a wide variety of supercapacitors on the market in the range of farads – from 0.1 F to 5 F or so – suitable for IoT sensor nodes. Products that lie within this range, and which offer leakage currents on the order of 10 µA are available from suppliers such as NessCap, AVX, Elna, Kemet, Nichicon, Panasonic Electronic Components, Seiko Instruments and Taiyo Yuden.
By exploiting the ability of supercapacitors to support many cycles and high instantaneous currents, it is possible to extend the lifetime of IoT sensor nodes and maintain their energy autonomy.
References
- Palma, L. et al, An Approach to Improve Battery Runtime in Mobile Applications with Supercapacitors, IEEE Power Electronics Specialists Conference (2003)
- Tanevski, M. et al, Power Supply Energy Optimization for Ultra Low Power Wireless Sensor Nodes, IEEE Sensors Applications Symposium (2013)
