Energy Harvesting for Low-Power Wireless Sensor Nodes

Today, wireless sensor nodes (WSNs) are finding a wide range of consumer, medical, and industrial applications. A jogger can now wear a Bluetooth-equipped chest band that can communicate her heart rate to a treadmill or wirelessly upload an entire workout from her PC and guide her through it via her iPhone. As the jogger ages, she can wear a wireless electrocardiograph (ECG) patch to relay her cardiac activity to a doctor.

On the way home from the gym, our jogger drives over a bridge whose condition is monitored by a wireless sensor network that monitors strain, cracking, and wear. Flying overhead is a plane whose wings are constantly monitored for mechanical stress by a network of wireless sensors powered by piezoelectric energy harvesters. She stops briefly at a gas station because a wireless tire pressure monitor (TPM) indicated that a tire was low. Arriving home, her ZigBee-equipped smart meter has been monitoring her household energy usage, providing feedback that helps her cut her electric bill.

All of these applications involve wireless sensor nodes that – with the possible exception of the smart meter – would do well to run for years between battery replacements. This can be achieved only by the use of energy harvesting, utilizing ambient sources to prolong the life of the batteries in wireless sensor nodes.

In this article we will examine the various ambient power sources available for energy harvesting and evaluate whether they are likely to prove adequate to power low-power wireless sensor nodes.

Energy harvesting power sources

Depending on the application and location, there are four potential sources available for energy harvesting: light, vibration, heat, and RF. We’ll discuss each in turn.

Light
In many, if not most applications, energy harvesting from ambient light will be an obvious choice. Outdoors during the day, solar energy flux averages 0.1 W/cm², which is sufficiently large that even a very small photovoltaic (PV) cell is usually sufficient to power a wireless sensor node.

Indoor solar sensors are frequently employed to manage lighting levels. However they are much less useful for powering wireless devices, since the power available from indoor lighting is typically 10 to 100 µW/cm², and its usefulness depends in part on the spectral composition of the light.

Sanyo Energy makes an extensive line of solar cells suited to a variety of applications. Their AM-1801CA is a 53 x 25 mm thin-film solar cell on a glass substrate designed for indoor use. Under maximum illumination it can output 4.9 V with a short circuit current (I_SC) of 20 µA. With less than 200 lux of fluorescent illumination, the AM-1801 can output 3.0 V at 18.5 µA.

For outdoor applications, the more powerful AM-5608CAR (see Figure 1) is a 60 x 41 mm amorphous silicon PV cell that outputs 5.1 V with an I_SC of 17.8 mA, more than enough to power a low-power wireless node as long as you store the excess energy for nighttime use.

Vibration
There are three basic approaches for converting vibration into electricity: piezoelectric, electromagnetic, and electrostatic. In piezoelectric transducers, vibrations deform the crystalline structure of the sensor, thereby generating electricity. In electromagnetic transducers, electricity is generated by the relative motion of a coil and a magnet. Electrostatic transducers respond to changes caused by vibration in the distance between two electrodes of a polarized capacitor. Most such energy harvesting devices developed to date are resonance frequency-based, generally involving the motion of a tuned cantilever arm to maximize the vibrational effect. Vibrational energy harvesters typically reach power levels ranging from a few microwatts of several milliwatts.

Midé Technology Corporation makes the Volture™ line of piezoelectric energy harvesters. Each Volture product uses a resonant beam of piezoelectric material that is clamped at one end and tuned to the resonant frequency of the vibration source for maximum output. While output depends heavily on both the intensity and frequency of the vibrations, a Volture V25W tuned to 40 Hz with a 15.6 gram tip mass on the cantilever and a 1 G vibration source achieves an output voltage of 9 V and output power of 9.2 mW. While that is a peak value, a steady output of 2 to 3 mW should be enough to help power most wireless nodes for a considerable amount of time. Figure 2 shows how Volture can be used to power a wireless sensor node.

Heat
Thermal energy harvesting is based on the Seebeck effect. Two dissimilar metals kept at different temperatures will display an open circuit voltage between them, with the voltage varying directly with the temperature differential. If the Seebeck coefficients of the two metals remain relatively constant over temperature, the available voltage is approximately

V = ΔS * ΔT

where ΔS is the difference in the Seebeck coefficients of each metal and ΔT is the temperature difference between them.

The simplest thermoelectric generator is the thermocouple, a junction of two dissimilar metals. Most thermoelectric generators (TEGs) are thermopiles, consisting of a large number of thermocouples connected in series and sandwiched between two metal or ceramic plates.

The amount of voltage and, therefore, power available from a thermocouple varies directly with the difference in temperature between the two metals. In practice, maintaining a high enough temperature gradient to produce useful energy can be a challenge due to the relatively low efficiency of thermocouples. For human body applications (T=22° C, ΔT≈15° K), the efficiency can be as low as 0.8 percent, and for industrial applications (T=22° C, ΔT≈50° K) only 2.7 percent. Still, pack enough small thermocouples tightly together in a TEG and you can get quite useful output. TEGs can typically generate anywhere from 2.8 mW when ΔT =10° K to 18 mW when ΔT= 30° K.

Laird Technologies - Thermal Division makes an extensive series of thermoelectric modules (TEMs) for energy harvesting applications. Their 15 x 15 mm² CP10,31,05 modules (see Figure 3) can generate up to 8.2 W at 25° C with a ΔT_max of 67° C. Clearly designed to fit between a power transistor and its heatsink, even if your wireless application cannot possibly generate enough heat to ramp these modules to full output, they deserve a serious look as a supplemental power source for any energy harvesting application.

RF
Even if your local TV station puts out a powerful signal, capturing enough ambient RF energy from it to power even the smallest device is a good trick, though it can be done.

The hurdle you need to clear for ambient RF harvesting is the Friis equation:



Basically, you should have a powerful transmitter or a good receiving antenna – preferably both – to be able to capture a useful amount of RF energy. Intel conducted one experiment using an antenna of 20 x 30 cm to capture the signal of a local TV station 4 km away. They were able to harvest 60 µW of energy, which translates to an energy density of 0.1 µW/cm². In urban areas, ambient RF might occasionally prove to be a useful supplemental energy source, but its usefulness will be highly location dependent.

Power management

Generating small amounts of power is one thing, managing it is quite another. The micropower sources used in energy harvesting present some unique challenges for energy management.

Piezoelectric sources
Linear Technology’s LTC3588-1 power management IC (PMIC) integrates a low-loss, full-wave bridge rectifier with a high-efficiency buck converter to form a complete energy harvesting solution optimized for high-output impedance energy sources such as piezoelectric transducers. The LTC3588 rectifies the voltage waveform and stores harvested energy in an external capacitor, bleeds off any excess power via an internal shunt regulator, and maintains a regulated output voltage by means of a nanopower high-efficiency synchronous buck regulator. The LTC3588 provides pin-selectable output voltages of 1.8 V, 2.5 V, 3.3 V, and 3.6 V at up to 100 mA output current.

The LTC 3588 is particularly well-suited for low-power wireless sensor nodes, since it accumulates energy over a long period of time to enable efficient use for short power bursts, such as those that occur when a node wakes up and transmits a burst of data. The frequency and power of the bursts can be adjusted to be directly proportional to the power available from the energy harvesting source.

Photovoltaic sources
PV cells present unique power management challenges. As the amount of irradiance increases, the current output from a PV cell increases far faster than the voltage output, which tends to plateau rather quickly. PV cells are basically current sources and need to be treated as such.

STMicroelectronics’ SPV1040 is a solar battery charger PMIC with embedded maximum power point tracking (MPPT). The MPPT is the operating point on an I-V curve where power output is highest. The maximum power point is located on the knee of the I-V curve and is the highest efficiency operating point for a PV device for given conditions of solar radiance and cell temperature. Every point on I-V curve is determined by the electrical load on the system. Since loads may be constantly changing, as will available radiance or lighting, some sort of MPPT tracking is desirable to dynamically match the electrical loads to PV output in order to maximize their performance. This is particularly true in the case of low-power wireless sensor nodes, where the current required for transmission will be many times higher than the quiescent current.

The SPV1040 is a low-power, low voltage, monolithic step-up converter with an input voltage range from 0.3 V to 5.5 V. The SPV1040 uses a 100 kHz PWM whose duty cycle is controlled by the MPPT algorithm to maximize the energy generated by even a single solar cell. The SPV1040 stops the PWM switching if either the maximum current threshold (up to 1.8 A) is reached or the maximum temperature limit (155° C) is exceeded. STMicroelectronics also offers the STEVAL-ISV006V2 evaluation board (see Figure 4) that enables you to check out the SPV1040 in action.

Energy storage

Another challenge with micropower energy sources is that their energy is often only intermittently available, so some method of storing excess power to match supply and demand is almost always necessary.

Batteries are the obvious storage medium, ranging from the ubiquitous CR2032 coin cell to large sealed lead acid batteries. DigiKey carries an extensive line of batteries, including lead acid, lithium, NiCad, NiMH, and thin-film. To get the maximum benefit from energy harvesting, choose a rechargeable battery whose durability will not be the limiting factor in the design life of your device.

Just as everything else electronic has been shrinking, so too have batteries. Cymbet’s solid-state, thin-film EnerChip™ batteries are increasingly paired with small PV cells in energy harvesting applications. The tiny (5.0 x 5.0 mm) CBC012 battery is rated at 3.8 V and can deliver 12 µA hours at a constant 50 µA drain. Texas Instruments and Microchip Technology use EnerChip thin-film batteries on their energy harvesting evaluation boards.

Thin-film batteries are usually paired with either a large capacitor or a supercap in order to be able to handle the current surge when a wireless node transmits. Ultra-high-density supercaps are an obvious choice as energy buffers in energy harvesting applications. Taiyo Yuden’s PAS414HR-VA5R is characteristically extremely dense compared to the same size electrolytic. The 60 µF/3.3 V surface-mount device is only 4.8 mm in diameter and 1.55 mm high.

Unlike batteries, supercaps show excellent cycle life and no issues relating to overcharge and over discharge. If your energy harvesting source is adequate to meet the requirements of your wireless sensor node, then a sufficiently large supercap may eliminate the need for a battery altogether.

Try before you buy

There are a number of very complete development kits that will enable you to evaluate different energy harvesting tools and incorporate them into your next wireless design.

Texas instruments’ eZ430-RF2500-SEH solar energy harvesting development tool (see Figure 5) is a complete solar energy harvesting development kit designed to help create a perpetually powered wireless sensor network based on the ultra-low-power MSP430 microcontroller. The solar energy harvesting module includes a high-efficiency 2.25 x 2.25 inch solar panel optimized for operating indoors under low-intensity florescent lights, which provides enough power to run a wireless sensor application with no additional batteries. Inputs are also available for external energy harvesters such as thermal, piezoelectric, or another solar panel. The system manages and stores additional energy in a pair of thin-film rechargeable EnerChips, which are capable of delivering enough power for over 400 transmissions.

Microchip Technology makes the DV164133 energy harvesting development kit based on the XLP PIC24F16KA102 microcontroller . Powered only by light, the XLP kit enables rapid prototyping of low-power applications such as RF sensors, temperature/environmental sensors, utility meters, remote controls, and security sensors. The board features an expansion PICtail™ connector with MC-controlled power supply so you can add RF connectivity with MiWi™-based low-power communication software and the MRF24J40 (802.15.4) 2.4 GHz radio.

Silicon Labs offers its ENERGY-HARVEST-RD, a complete reference design board built around a solar cell and their EZRadioPRO® chip (see Figure 6). The system consists of two components: a wireless sensor node and an EZRadioPRO USB dongle. The sensor node uses a Silicon Labs’ Si1012 wireless MCU; the dongle uses a Silicon Labs’ C8051F342 MCU and a Silicon Labs’ Si4431 radio. The sensor node operates at 919.84 MHz and is powered by a solar energy harvesting power supply. Since it is powered from an energy harvesting source, no batteries need to be replaced for the life of the system (life expectancy is greater than 15 years or 7000 mAh).

Is energy harvesting enough?

Depending on the application and the availability of potential ambient energy sources, using energy harvesting techniques to power a wireless sensor node can make a great deal of sense. Table 1 summarizes the amount of power available from each of these different sources, which will factor heavily in determining your power budget.

Source	Source Power	Harvested Power
Light
Indoor	0.1 mW/cm²	10 μW/cm²
Outdoor	100 mW/cm²	10 mW/cm²
Vibration/Motion
Human	0.5 m at 1 Hz
	1 m/s² at 50 Hz	4 μW/cm²
Machine	1 m at 5 Hz
	10 m/s² at 1 kHz	100 μW/cm²
Thermal
Human	20 mW/cm²	30 μW/cm²
Machine	100 mW/cm²	1-10 mW/cm²
RF
GSM BSS	0.3 μW/cm²	0.1 μW/cm²

With the exception of RF, available energy harvesting devices can typically supply 10 µW to 1 mW, though rarely on a steady basis. Even assuming that you can store and meter out the power supplied by an energy harvesting source, whether or not any one source is accurate will require looking at both the power and usage profiles of your application as well as the power that can be derived from it and the energy harvesting source in your proposed application over a typical 24 hour period. However you look at it, some sort of energy storage system – whether a rechargeable battery or a super capacitor – will be required.

Depending on both the application and location, you might do well to combine different ambient energy sources. For example, an outdoor wireless sensor node might well use a solar panel with a bank of TEG's mounted on the back of the panel and connected to a heat sink below. In this way, the heat generated in the solar panel by the sun can be drained away and converted to useful energy. This has the additional benefit of lowering the operating temperature of the PV array and thereby increasing its efficiency.

Similarly, a wireless sensor node set up to monitor bearing wear in a generator would certainly use a piezoelectric source to convert motor vibration to useful energy; at the same time a TEG mounted near a hotspot on the equipment could provide additional energy.

Looking ahead

The bottom line is that low-power wireless sensor nodes really require some form of energy harvesting – whether from single or multiple sources – in order to minimize maintenance and extend the life of the devices. As wireless sensor networks become more widely deployed, energy harvesting will no longer be an afterthought, it will be a central part of every design.