By: Cristian Toma, Applications Engineer, Vivien Delport Applications Director, Microchip Technology Inc.
Smart environments rely on a variety of sensors, controllers and actuators, which play a number of roles and are distributed across the entire environment. This distribution creates some technical challenges. For example, each sensor requires its own power source. Monitoring a low-battery condition is a standard operation. However, replacing the battery requires human assistance. This paper proposes the implementation of an energy-harvesting, low-power sensor. Such a sensor can be totally maintenance-free for a period of several years, when energy harvesting is employed, while a battery-powered sensor would run out of power within a few months.
Various implementations of wireless sensors are available today. The overall cost of such a system depends on more than just the hardware. The cost of implementing different industry standards can add to the overall cost.
This paper proposes some simple low-power, energy-harvesting techniques that can be used to implement maintenance-free wireless sensors. Additionally, this paper will show how to keep the overall cost low while providing solid performance.
An energy-harvesting system stores energy for use later when needed. The main difference between energy-harvesting wireless sensor and battery-powered sensor is that the later is designed to operate with that battery for a specific period of time. An energy harvesting sensor node has the advantage that it can harvest energy indefinitely for later use. Typically, the amounts or energy that it can harvest are very limited. So, the energy usage from the wireless transmitter and the sensor itself must be balanced, in order to use less energy that the harvesting section can provide.
Different Energy-Harvesting Devices
There are different kinds of energy harvesting devices available in the market. The most common used device is solar panels. They come in different sizes, from the large panels with many solar cells grouped in series and/or parallel, to the very small cells used to power handheld calculators or toys.
Another type is the RF harvesting device. This device receives radio waves using an antenna, and converts them into electrical energy. This is a different type of energy-harvesting device as it requires high levels of RF energy. Electro-mechanical harvesting devices typically use a moving magnetic part in the vicinity of an inductor coil. Thermo-electrical energy harvesting devices can produce small amounts of electrical energy from temperature gradients. These thermopower devices rely on the Seebeck effect.
Implementing a Wireless Standard (or not)
When adding wireless capabilities, some users tend to only think of implementing RF industry standards such as ZigBee or Bluetooth. However, depending on the actual application requirements, implementing a specific standard may or may not be a real requirement. Implementing a specific standard is only needed when the end product is compatible with products already existing on the market. The choice of making a product compatible with other products is really a more complex business decision. There are tradeoffs on whether to provide compatibility. Also, there can be cases when compatibility is a must, just as there are cases when adding compatibility is not possible or becomes too expensive.
Additional Costs of Implementing a Standard
When looking at implementing a specific RF standard, the designer is only looking at the overall hardware cost. This is generally the main starting point, when looking at a hardware solution. Any RF transmitter (officially called “intentional radiators”) needs to be certified. Non- RF transmitters still need FCC or CE certification. For any wireless sensor, the FCC certification is unavoidable. So, when comparing different solutions, this cost component will be left aside.
Depending on the implemented wireless standard, the overall cost of implementation can be much larger than originally anticipated. These costs typically include items such as group membership, standard compliance testing, specific profile testing, specific hardware sniffer tools, etc. The cost of a ZigBee certification is around $3000. This is just for the certification itself. In practice, before applying for certification, one needs to do some specific pre-tests to estimate the device will pass the certification. Specialty test equipment can be rented at $750 per month.
These additional costs might not seem high. However, many times adopting a specific standard also requires you to pay a membership cost. There can also be royalties you need to pay. The RF standard certification cost will always translate to additional cost and additional delays, until a product is released.
The per-unit-cost of the hardware itself is typically between $1 and $1.5 each for 10k units. All of the costs will impact the overall cost per unit when producing low quantities. If we only take into consideration the FCC certification cost of $10,000 then this will result in effectively doubling the price per unit. The cost of RF standard certification can easily go beyond $10,000.
Minimum Hardware Requirements
A specific wireless standard would require the use of a specialized chip. However, if you only need one-way communication, a simple ISM-band transmitter will suit the application. There are some minimum requirements for an energy-harvesting wireless sensor node. The use of high data rate is recommended. A higher data rate also requires a bit more power. But the overall packet length is much smaller, thus consuming less energy. The modulation used can be either ASK (OOK) or FSK. ASK modulation (and OOK) is using less energy because it has periods when the RF power is smaller (or none at all for OOK). Overall average current consumption for ASK will be lower. Still, FSK is preferred because it can go to substantially higher data rates. For example, the PIC12LF1840T48A MCU with integrated transmitter from Microchip supports 10 kbps in OOK mode and 100 kbps in FSK mode. In this case, when using FSK modulation, the data can be sent ten times faster. From the RF receiver standpoint – receivers can receive and decode FSK signals much better than ASK modulated RF, especially at higher data rates.
Optimizing Power Consumption
A wireless energy-harvesting sensor needs to use as little power as possible while in operation. This can be achieved by balancing the active periods with the low-power shutdown modes of the device. Depending on the response time of the application itself, the sensor will need to transmit more or less frequently the measured sensor information. The longer the time between two active periods, the lower the average power consumption, and lower the actual energy usage.
The sensor also need to capture multiple data samples between two radio transmissions. Depending on the actual physical information that is captures, this can draw more or less current. Example includes op amps and a bridge load cell that requires relatively large amounts of current while active.
Attention needs to be placed on the actual wireless RF transmission configuration. Parameters like amplitude or frequency modulation, the speed at which the information is being sent (bitrate and/or frequency deviation) and the RF output power into the antenna all have an impact on the overall power consumption. The system needs to be carefully designed to eliminate all unnecessary power usage. The processor must stay in a low-power state as long as possible. All other on-board devices must be in a low-power standby mode when not used.
The RF transmitter on the PIC12LF1840T48A device features a maximum frequency deviation of up to 200 kHz. This will allow a maximum bitrate of 100 Kbps. If we use a small data packet consisting of a 16-bit preamble, a 16-bit synchronization pattern and a 32-bit payload it will only take 640 μS to transmit one complete data packet. By knowing that the energy is measured in units called joules (J) and that:
1J = 1W * 1s = 1V * 1A * 1s
We can easily calculate the energy consumption used for sending one data package as:
E = 10.5mA * 640μS -> 10.5mA * 3.0v * 640μS =
31.5mW * 640uS = 20.16μJ
For our PIC12LF1840T48A design example we know the crystal oscillator startup time is typical 650 μS and it draws around 5mA when it starts-up the crystal oscillator. This gives us a startup power consumption of:
E1 = 5mA * 3.0v * 650μs = 9.75 μJ
The actual data transmission used in our example contains 16 bits of preamble (101010….), 16 bits of synchronization pattern and 32 bits of data. For the selected bit rate of 100 kbps, this will give us a transmission period of 640 μs640μs. For a +0 dB RF transmission at 868 MHz, FSK modulated we have a power consumption of 12 mA.
E2 = 12mA * 3v * 640μs = 23.04μJ.
I we would have used a simple 10kbps transmission then the energy used would have been:
E2 = 7.5mA * 3v * 6.40ms = 144μJ.
This comparison is just to outline the importance of using a higher data rate.
After sending the last data bit, the PIC12F1840T48A transmitter will automatically time out and go back to a low-power shutdown state. This timeout period has a minimum value of 2 mS. This will give an additional energy consumption of:
E3 = 12mA * 3v * 2ms = 72μJ.
Thus, the total power drawn to transmit one data packet is:
E = E1 + E2 + E3 = 9.75μJ + 23.04μJ + 72μJ =104.79μJ.
A miniature solar cell that generates a current output of 4.5 μA @ 3V will need to be active for however many seconds it takes to get enough energy for only one data transmission. For example, using a low-cost solar cell that produce 3V @ 6mA,(3v @ 40μA, (best case) only gives us generated power of:
3v * 40μA = 140μW
We can now calculate how long we need to collect energy to send one single data transmission:
T = 104.79μJ/ 140μW = 0.74s
This means that the sensor unit must wait a little less than one second between two sequential data transmissions. Care must also be taken, as the above calculation was made for the case when the solar cell has a constant light source indefinitely. Of course, this is not true for the most common case, where the primary energy source is the natural light that is available only during the day. In this case, the calculation must be expanded so it will take into consideration that during daytime the harvesting system needs to store energy for the nighttime, when there is no natural light. Also, the energy required by the actual sensor measurement was not calculated in this example.
Possible Implementation Options
Depending on the actual system requirements, there are multiple energy storage options in implementing the energy-harvesting feature.
– Harvesting energy into a super capacitor.
– Rechargeable battery. NiMH rechargeable batteries can be trickle charged directly from the solar cell. There is no need for a charging regulator or a specific charging regulator. Also, NiMH rechargeable batteries have a very low cost.
– Powered directly from the energy harvester. In cases where the primary source of energy (such as light or heat) is continuously available and the generated power is enough to supply the wire less sensor circuit, there is no need to store the energy into a separate device. Of course, the applicability of this option is very limited.
Why energy harvesting
The primary benefit of using energy harvesting solutions, when developing low-power wireless sensor nodes, is in the cost savings in the deployment and maintenance of the wireless sensor system. The maintenance cost in monitoring and replacing batteries of wireless sensor networks can easily overshadow the per-unit cost savings, especially if the wireless sensor system is installed in remote or difficult to reach areas. The size (number of sensors) of the wireless system also becomes a bigger factor when regular maintenance services are required. Energy harvesting technologies allows us to collect “free” energy and store that energy for when we really need it.
Energy-harvesting wireless sensor nodes can be designed at more competitive price points these days. Most of these new wireless sensor designs do not even need a battery and can harvest energy from different primary energy sources, such as light, radio waves, mechanical energy and heat.
In normal conditions, an energy-harvesting low-power wireless sensor can function almost indefinitely and never require any human intervention. This can be a significant maintenance cost saver, especially when the sensor is located in places where human access is difficult. By being more careful when selecting communication protocols, transmission data rates, and better utilizing power management features found on new RF devices, like the new PIC12F1840T48A, we can significantly reduce the overall power requirements and also the cost of your wireless sensor solution.