Visit US site

Energy Harvesting for M2M Wireless Networks

Fig1The concept of harvesting ambient energy from the immediate environment is not new; wind and water have been such sources for hundreds of years. In contrast the concept of making a wireless, batteryless solution has only recently been achievable due to recent advances in energy efficient electronics.

Using Energy Harvesting devices as end nodes is ideal for a number of M2M wireless networks where, typically, there is an 'always on' device being used such as a GPRS or internet gateway. See Fig1.

M2M wireless networks benefit from energy harvesting technology as it eliminates the need for battery replacement, making them ideal  in hostile or inaccessible areas .They give a step change in longevity and are also environmentally friendly as no toxic chemicals are used or disposed of.

The principle energy sources available in a typical M2M environment are kinetic energy, light and heat. Harvesting these sources of low energy poses a problem if you want to maximise wireless range. Power is ratio of energy over time and some devices, like kinetic harvesting switches, do not operate over long periods! - Fast and reliable transmission protocols are vital and these have been incorporated with good effect into the EnOcean Wireless standard.

Energy Harvesting Devices

Fig2 - inductive harvester used in the PTM200

Harnessing Energy from motion such as the action of pressing a device is one possible solution, but finding a suitable and sustainable method to extract energy is no simple task. Piezoelectric and inductive generators are the most common. Piezoelectric devices can give a small footprint per joule of energy but can also be inefficient and mechanically unstable.

Inductive solutions are larger, lower cost, have excellent reliability and are more efficient. When using an inductive system as a switch there is little mechanical movement and only a few Newtons of force in the switching action.

So some sophisticated techniques are needed to generate enough energy to power a wireless switch, which can be achieved in the harvester design (Fig2) that is used in the EnOcean PTM200 switch; With sufficient care it is possible to design a product in combination with the electronics to transmit 3 data packets or telegrams per button push allowing both on / off and dimming commands to be transmitted

Light

Fig 3 Power available from low cost thin film solar cellsSolar cells are an ideal energy source for most areas and a small 8 cell design can deliver 11-14uA at 3-4V (Fig3).

Light outputs are usually variable and of course can be absent, suitable energy management and storage schemes need to be employed to avoid shutting down devices.

Whilst NiCd or Lithium batteries are low cost they do not tend to last too long and require complex charge circuits and replacement after a certain number of charge cycles. If a maintenance free system is required a better approach is to use a PAS. A PAS or PolyAcenic Semiconductor is smaller than a double layer capacitor and has a high capacity per mm2 with low self discharge rates. Also it's environmentally friendly having no Cadmium, Mercury or Lead. A PAS coupled with an effective energy management scheme makes Battery less wireless nodes such as PIRs, Thermostats and CO2 sensors simple to implement at fairly low cost.

Heat differential can be utilised in many locations as a source of power for a range of remote applications.

Fig 4 Seebeck EffectA thermoelectric device creates a current when there is a different temperature on 2 junctions of 2 metals, a property discovered by Seebeck in 1821. Referring to Fig 4 , a heat source will drive electrons in the n-type element toward the cooler region, creating a current through the circuit. Holes in the p-type element will then flow in the direction of the current. The current can then be used to power a small load.  Conversely when a voltage is applied, it creates a temperature difference (known as the Peltier effect). There are a number of low cost Peltier elements available and these can be used 'in reverse' as generators as shown in fig 4 for small wireless monitors. The voltage generated from these elements is, however, very low and requires innovative DC-DC conversion to get the voltage to a level that can be used by a typical wireless controller

A suitable device for this purpose is the ECT310, a low-cost ultra low voltage DC/DC converter that uses an innovative blocking oscillator design.  Wireless sensors and even actuators can be powered by this method given suitable charge management.

Fig 5 Peltier powered wireless node

With a 2 Kelvin temperature difference and a standard low-cost Peltier element the DC/DC converter starts operating at around 20 mV giving an output depending on the actual temperature difference of the Peltier element.  An input voltage range of 20 mV to 50 mV corresponds to an output voltage range between 3 V to 4 V. A typical thermo-driven sensor consists of a sensor element, a small Peltier element, a DC/DC converter and a radio module as shown in Fig5.

Fig 6 shows the block diagram of a complete sensor node. Implementing a wireless system using harvested energy constrains microcontroller (MC) design so a balanced approach is required to ensure energy is available for sensing control.

Fig 6 . A typical Energy harvesting sensor nodeThe start-up time of a microcontroller plays a very important role and is usually influenced by oscillator delay. Crystals and ceramic resonators can take several milliseconds to stabilize. RC oscillators, by contrast, provide fast start-up but generally suffer from poor accuracy over temperature and supply voltage. To save time it is advisable to use a microcontroller that can start with an RC oscillator and subsequently switch to a crystal oscillator.

Fig7Rapid switching of a sensor (Fig7), rather than continuous operation, saves energy. This approach is particularly effective when measuring parameters that change slowly .It's possible to achieve an average current consumption just above the total current consumption of the continuously running processor blocks given suitable power strategies

While several circuit blocks can be switched off, others, such as threshold switches, must be operated continuously; which activate electronics and timers that trigger periodic activities such as sensor readings. These circuit blocks rapidly dominate the entire energy requirements and must be aggressively optimised. The timers of typical harvesting modules should require only approx. 20 nA, so these are typically analogue & switch off all components during sleep periods. This can enable a power reserve via the PAS of up to one week with solar devices, even in complete darkness.

If there are highly dynamic processes that need to be analyzed, it is worth pre-processing the data in the sensor and reduce the data to be transmitted by transmitting measurements only when there is a change from the last measurement. Simple well architected software performance reduces execution time saving more energy.  Another thing to consider when waking a CPU is the oscillator start-up and power down times, these include factors such as the time to enter and exit the mode and the energy consumed by doing this. Avoiding flash, EEPROM and other memory writes clearly saves power.

In order to conserve power it is necessary to employ a number of routines to balance power demands  - Besides an Off mode that clearly shuts everything down and one fully active CPU mode  a number of interim modes make sense for strategically using harvested power. For example four standby and sleep modes can use various timers and frequency sources allowing the user to select the best power savings and timing accuracy;

  1. Deep Sleep Mode - used for weak ambient energy powered, event triggered Transmit applications, where ultra low power consumption is mandatory. - only the ultra low power blocks are active.
  2. Flywheel Sleep Mode - used for high precision system timing in low duty-cycle networks. This timer has a programmable cycle time and is based on a wristwatch crystal oscillator.
  3. Short-term Sleep Mode - is used for interrupts which are significantly longer than the main crystal start up time (i.e. between RF subtelegrams) and is based on a Short Term RC Oscillator with moderate accuracy.
  4. Standby Mode - Timers stay running and all the content of registers and RAMs stay alive to ensure the fastest wakeup and highest timing accuracy in exchange for higher power consumption.

The Radio system

Fig8Power efficient design of the radio circuitry and protocols is critical to the successful implementation of an M2M wireless network powered by ambient energy, EnOcean's protocols and modules are a good example of this. The extremely short telegrams of EnOcean's wireless modules enable operation of a large number of transmitters in the same area. Error rate caused by telegram collisions remains extremely low because transmission reliability is better than 99.99% even with  100 wireless sensors where each node is transmitting data once a minute. Large office buildings or industrial plants can be equipped with high numbers of EnOcean sensors, and all will operate reliably and nearly simultaneously.

Implementing an Energy Harvesting M2M system

From the information presented above it's clear that there are constraints when using an Energy Harvesting node, the largest issue is that they are not typically 'on' much of the time.  They are also not suitable for long range communications.  These issues do not represent a problem for M2M systems as most have a higher power device that is communicating to a back office system or local PC/Server and is always active or at least powered.  We therefore have an ideal situation where end devices can run wirelessly with no maintenance or batteries but can communicate via an IP network via a gateway.

For example in a Smart Metering Home Area Network ( Fig9) the meter is always active and it will either communicate from an on board module , to an active hub or work with another device such as an ADSL / GPRS router communicating to a back office system via the internet/cloud.

Fig 9 Smart Metering HANA harvesting network can make good use of the always powered devices for communication, monitoring and control. In the simplest M2M system a harvesting switch can transmit to a receiver that has a triac or relay and this will behave as an on off / pwm control type system. This is ideal for lighting or control of valves etc. The receiver has all the 'intelligence' and will sense when to either behave as an on/ off switch or activate pwm mode. In an Enocean Switch there are 2 'signals'; one when pushed, the other when released. If the receiver does not 'hear' a release signal it assumes it must be dimming / pwm command.

Two way communications are important for many M2M systems and this can be easily achieved with more sophisticated Harvesting protocols. The main challenge in an energy harvesting system is to balance use with the limited available energy. This concept has to address a receive mode that consumes a relatively high amount of energy together with unknown delay times introduced through repeaters.

A concept called Smart Acknowledge uses a mechanism where the sensor expects to receive a radio message from the controller in a predefined time slot. To achieve this, a sensor sends a request to the controller from which it is expecting to receive data (reclaim) and then turns on its receiver for a short period of time.

In order to eliminate any latency, the return data is delivered to a 'mail box'. This mail box needs to be located on a powered transceiver with direct radio connection to the sensor. This transceiver can be the controller itself if no repeater is required, or a repeater which has a direct radio connection to the sensor. The transceiver that administrates the mail box is called a post master.

Fig10a

Fig 10a shows a setup where the sensor has direct radio contact to Controller.

Fig10b

Fig 10b shows a setup with a repeater acting as post master between the controller and the sensor, and highlights the difference between the actual radio link and the logical connection.

For a SMART ACK system to work, it is necessary to establish the relationship between the sensor and the controller and to define who is acting as the post master and his is done during the learn process. Once the system is taught it can exchange the required data during normal operation. More details on this can be found at www.enocean.com

Fig11A  complete M2M system can have a gateway communicating with local area controllers that talk to energy harvesting end nodes. This gives connectivity to the internet as illustrated in fig11

What's already out there:-

If we consider the Enocean Standard protocol there have been over 200,000 projects 'enabled by EnOcean' since 2003. There are also over 750 interoperable products available from over 100 product manufacturers.  Additionally there are Interfaces to established automation solutions such as MODBUS, LON, EIB/KNX, BACNet TCP/IP that can form the backbone of an M2M wireless network.

Devices available include switches, temperature controllers CO2 sensors transceivers and gateways and can be seen at -  http://www.enocean-alliance.org/en/. Many of the devices are ideal for Home Area networks working with smart grid and smart metering networks.

In the future other remote sensing devices are being planned including:-

  • Rotary harvesters that can be powered and monitor water flow.
  • Solenoid and ball valve controllers
  • Structural monitoring systems
  • End to End Smart Metering systems

Details of all enocean module together with more in depth application information can be found at - www.enocean.com

Last updated: 17 October 2011