Internet of Things deployments are taxing on many technology facets, perhaps none more so than the types of batteries...
they use. For example, Z-Wave door locks can run for two years on AA batteries. In a connected home, maintaining these locks every couple of years doesn't take much effort. However, when scaled to include hundreds -- if not thousands or millions -- of such locks that need batteries replacing, things can get out of control -- and fast.
To accommodate the Internet of Things and let enterprises truly benefit from the promises it brings, companies must carefully consider which IoT battery is used in their applications.
An inside look at IoT battery options
Chemical batteries -- lithium, alkaline or lead acid -- convert chemical energy to electrical energy. A chemical reaction causes electrical ions to flow from the negative terminal (anode) to the positive terminal (cathode) of the battery through some kind of electrolyte. To recharge the battery, that flow is reversed and thus converts electricity back to a chemical form.
Lithium is the lightest metal and has the greatest power per weight ratio (called energy density, expressed in watts per hour per gram) of any chemical battery, thus making it suitable for items that are small, such as watches and IoT sensors. However, lithium-based batteries have some inherent problems, the major one being that lithium is volatile and can catch fire.
Alkaline batteries are a much older technology. They are the round, cylindrical batteries you have likely used for years. The electrolyte in an alkaline battery is manganese dioxide.
Finally, there are lead acid batteries, such as those used in cars. A 12-volt car battery has six lead acid batteries inside, each producing 2 volts.
To understand different types of batteries and power usage, it is important to know how they are measured. The relevant metrics are current (ampere or amps), voltage (volts) and power (watts). Algebraically this is:
amps = watts / volts
volts = watts / amps
watts = volts * amps
Think of amps as load, which is measured in terms of electrical resistance (ohms or Ω). Another way to look at load is if you drive an electric motorcycle uphill, it will run down the battery faster than if you drive it on flat ground. To say that it "runs down" means its voltage falls off not to zero, but to a level that it can no longer provide enough volts to power whatever it is supposed to power. The measurement of how much energy a battery has stored is called charge, measured in milliamp hours (mAH).
Current IoT battery options
A few batteries are commonly used in small devices, including lithium, nickel and alkaline.
Consider the popular CR2477 lithium disposable coin cell battery. This battery has 1,000 mAh capacity and costs about $3.60 each. A typical motion detector sensor, which runs on 3 volts, uses about 20 mA at peak power usage, so in theory, if you rocked it back and forth continuously it would last for 1,000/20 = 50 hours. This is a simple explanation. There are other factors that affect how fast a battery runs down, such as temperature.
Nickel cadmium batteries were for a long time the way to power portable radios and computers, however have since been losing market share to nickel metal hybrid (NiMH) batteries. NiMH has greater power density than nickel cadmium. NiMH batteries are often used in camera flashes as they are suitable for high-energy applications and last longer than alkaline batteries under a heavy load. However, NiMH batteries can be damaged at sub-zero Fahrenheit temperatures. Additionally, both types of nickel batteries also subject to what is called the memory effect, meaning they retain less charge unless you run their voltages down to as low as possible before recharging them.
Alkaline batteries are suitable to multiple usages as they come in all sizes, including A, AA, AAA, D and so forth. However, they are not rechargeable like nickel or lithium batteries.
Future IoT battery options and alternatives
Given the relatively high price of lithium batteries and the heavy weight of alkaline and lead acid batteries -- and given that all of these battery technologies are old -- engineers and physicists are coming up with new and improved IoT battery options. Most of these have not yet matured into products on the market yet, for different reasons.
Toyota is researching a sulfide superionic conductor battery that can charge in seven minutes. Researchers in South Korea are working on stainless steel batteries that could power a cell phone for a week. Researchers are also experimenting with other metals and chemicals such as sodium ion and copper foam substrate. Additionally, there are solid state batteries that replace liquid lithium ion with graphite. Researchers are also looking at ways to draw power from dew, human skin (motion) or using sound to transfer power. The goal of this research is to both find an electrolyte that is less expensive and rare than lithium. This would extend life by allowing more energy per weight; a smaller battery could do the same job, or keep the same size and have it last longer.
There are also flexible options such as thin-film solid-state batteries. The advantage with these is they could, for example, be wrapped around a wrist to power a health monitor. A 3D-printed zinc battery is suitable for wearable devices as well; startup Imprint Energy is already manufacturing them.
There are also alternatives to the chemical-energy approach to storing electricity. A capacitor, which is named for its capacity to hold an electrical charge, holds static electricity. While the capacitor is an old idea, the supercapacitor is new. The appeal of the supercapacitor is that you can charge one rapidly, but they have not gained much traction as they cost a lot, weigh a lot and do not hold a charge long, losing up to 25% of their charge per day. The Maxwell D 2.7 v costs about $20 and weighs 60 grams.
Another option is ambient energy harvesting. There are a lot of proposed ideas, but scarcely any in production because of the technical issues involved in both reducing the size of the device and, equivalently, boosting its energy conversion ratio. One example, proposed by EnOcean, is to draw power from the temperature gradient of food containers to phone home where there is a risk of spoilage.
Physicists have long known that you can generate power from light, motion, temperature differentials and electromagnetic radiation. Solar panels, for example, use the photovoltaic effect, which captures light from the sun. Motion, such as vibration, also generates power. You could essentially generate power from a bridge by capturing the energy as cars drive over it. Temperature differentials generate power too, so, you could, for example, use an energy harvester next to an engine or water heater that is hotter than the air around it, thus producing the necessary temperature differentials to exploit the physical effect.
Radio waves can generate power as well. A Mouser power harvester, for example, can produce 5 volts of output from -11 dbM (decibel milliwatt) of radio power. A dbM is a fraction of a milliwatt, expressed as a logarithm. To understand how powerful -11 dbM is, consider that Bluetooth operates at 0 dbM which is 1 milliwatt (1/1000 watt). As I write this, my cell phone shows two bars and -85 dbM; -10 dBM is a strong Wi-Fi signal but fades away to where you cannot use it at all at -100 dBM. The actual transmitter on your cell phone outputs 21 dbM. The point here is that it does not take much power to run, say, a vibration sensor using radio waves.
Power over Wi-Fi
Since RF can be harvested, why not use Wi-Fi to power IoT? In 2015, University of Washington researchers demonstrated it is possible. However there are no products on the market yet, likely due to the difficulties in manufacturing these devices the researchers suggested in their paper as well as the need to modify both the Wi-Fi router and the IoT sensors.
The idea behind using electromagnetic radiation to transfer electricity over the air is not new; there are already wireless cell phone chargers. However, the problem with the current Wi-Fi specification is there isn't any power going back and forth between the Wi-Fi router and the device attached to it unless there is some traffic, like a person surfing the Web. So the researchers proposed the Power Over Wi-Fi standard and built a prototype of a smart home to show how it works. They changed the router to send a continuous signal, in turn creating constant traffic and thus a steady stream of power. They demonstrated that they could use Wi-Fi to charge nickel and lithium battery chargers at up to 28 feet away, and then powered a temperature sensor and a camera.
The good news is that all of this works using the transmitters and antennae that people already use: Bluetooth, Wi-Fi and ZigBee antennas. The bad news is their circuitry would need to be modified.
Understanding the power requirements of IoT
IoT devices generally do not use cellular or Wi-Fi as they would be too costly, and while they provide a lot of bandwidth, it would be overkill; a sensor might only need to send kilobytes of data. Wi-Fi HaLow is a new spec for Wi-Fi that aims to reduce the power needs of Wi-Fi by working in the sub-gigahertz range. Wi-Fi HaLow operates in the 900 MGz (0.9 ghz) range while regular Wi-Fi works at 2.4 to 2.5 Ghz. However, Wi-Fi HaLow is not ready yet, and no products are expected to reach the market until 2018.
The Z-Wave and ZigBee protocols, on the other hand, are designed to use little power. Z-Wave devices, such as a door or window sensor, motion sensor or smoke detector, do not need routers and don't connect to the Internet directly, instead connecting to something such as the Samsung SmartThings Hub to send data to an IoT application in the cloud. A sample Z-Wave transmitter uses 23 mA.
ZigBee uses multi-node mesh networking to relay its signal to a network gateway by hopping from one device to another; ZigBee devices also do not need routers, there just needs to be one at the end. ZigBee uses anywhere from 5 mA to 34 mA depending on the model.
Computing card, sensors and transmitters also affect power consumption. For example, this $45 GPS chip uses a low 25 mA, and this $50 Intel Edison computer card uses 100 mA when the radio is turned off. The Aruba BLE beacon uses two CR2477 batteries drawing a low 7 mA, which helps explain why beacons can run on batteries for two years.
Low-power IoT networks: The second wireless revolution