[ IoT 101: Power Supplies, Part 1 ]
You’ve designed the perfect weather station, with all the sensors you could possibly need. How are you going to power it? Unless you’re planning to run a 20-meter extension cord from the garage, you’ll need batteries! In today’s IoT 101 lesson*, we’ll cover everything you need to know about batteries for IoT devices, including how to select them and use them efficiently.
So many batteries
When was the last time you checked all the options in the battery aisle of your local electronics store? It’s hard to believe the world needs so many different types of batteries, but apparently we do. We all want the perfect battery for our electronics, and yet the perfect battery doesn’t exist today. So we make tradeoffs, and tradeoffs that made sense for somebody else’s product may not make sense for your IoT design. As a result, we end up with an aisle full of batteries that are certain to confuse or overwhelm your average consumer who thought he was just going to the store to pick up a pack of batteries.
Here are the primary considerations when choosing batteries for your IoT device:
- Single-use vs. rechargeable chemistry
- Cycle life (for rechargeable batteries)
- Energy density (how much power you can get per volume or per weight)
- Form factor
Do we have chemistry?
First, I need to cover some basic electro-chemistry (don’t worry, there won’t be a test). The power you need for your IoT device is a current (stream of electrons) delivered to your electronics at some voltage. For example, anARTIK 10 requires current supplied at between 3.4 and 5.0 V.
A battery generates electrons with a chemical reaction. The negative side (anode) of high-capacity batteries are typically an alkali metal such as lithium. The positive side (cathode) reacts with the anode to release electrons. Anode materials are selected for energy density, electron diffusion rates (max current), and thermal stability.
The state of discharge and chemical reaction determines the voltage across one cell of a battery; a cell of a lead-acid car battery ranges from 1.95 V to 2.10 V. You can stack cells to create a battery; 6 lead-acid cells equals a 12-V car battery. Use more lead and more acid to produce more current; consider the size of a 12-V motorcycle battery versus a 12-V truck battery.
If you want to use disposable batteries, then your users are going to look for “photo” batteries at their local store. These batteries are based on Lithium / Manganese Dioxide chemistry and are available in a variety of package sizes, including the CR123A cylinder (ANSI–5018LC, IEC-CR17345) and the CR 2032 button (ANSI 5004LC).
For applications using rechargeable batteries, Lithium-ion chemistries are the natural choice. You can source this chemistry in standard packages (including CR123A cylinders and 2032 buttons.) The most common type of Li-Ion battery uses lithium cobalt oxide (LiCoO2), which may pose safety problems when damaged. The size of battery pack required for an IoT device probably does not pose a safety hazard unless it’s a wearable device. In those cases you may want to consider alternative Lithium-ion chemistries that offer less energy density but higher safety.
For wearable devices you might consider a custom battery using Lithium-polymer chemistry. You won’t find them at your local electronics store, but they’re thin, flexible, and you can decide what shape fits in your device. Suppliers such as Powerstream offer flexible batteries as thin as 0.5 mm. And, you can find suppliers that offer “off the shelf” configurations for prototyping.
Time to get real
Enough theory! Let’s put some real batteries to use in a real project. Let’s power the weather station built with ARTIK 10 using CR123A batteries (Li/MnO2 chemistry).
In the real world, the voltage of a battery drops as it discharges. Here’s the discharge curve for an Energizer CR123A battery:
Notice two characteristics of the curve:
- It’s quite flat, so a large portion of the power is extracted before voltage drops significantly.
- It’s never high enough to power an ARTIK 10 device without a boost.
I can boost the output voltage two ways: by stacking two cells together to make a battery that stays above 4 V for most of its useful life, or by using a circuit that boosts the voltage. Voltage-boosting power supplies are more complicated than I want to implement for the weather station, so we’ll stack two cells to power our IoT node.
The ARTIK 10 requires a minimum of 3.4 V input power, meaning we can power our weather station down to about 2 V per cell. It’s obvious from the discharge curve that we will have extracted the vast majority of the power from our batteries before we drop below the ARTIK 10 threshold. (Our power regulation circuit will impose some voltage drop, but not enough to change the fundamental calculus that two CR123A batteries will work well.)
Reliable power regulation
Now we’re getting somewhere: we have a battery from which we can draw power, and an ARTIK 10 module that needs power. All we need is a circuit between them to regulate the power. The type of circuit we choose depends on how much we’re willing to spend to increase battery life.
Here’s the challenge: Let’s say we need to supply 200 mA @ 3.5 V from a battery with terminal voltage of 5.5 V. 200 mA leaving the batteries transfers 1100 mW of power, while 200 mA entering the ARTIK 10 module only transfers 700 mW of power. Our circuit needs to deal with the 400 mW difference.
The simplest and cheapest option is to dump the extra power by converting it to heat. That’s what a linear regulator does. A low-dropout linear regulator (LDO) allows the input voltage to drop very close to the output voltage before the regulator can’t operate. LDOs also provide very clean power, meaning tight regulation with very little ripple. The LDK220M36R from STMicroelectronics costs less than $0.25 in thousands, provides 200 mA at 3.6 V from sources between 3.95 V to 14 V, and requires no external components. You could increase battery life a bit by using an LDO that used an external component to trim output voltage to match the 3.4 V minimum required of the ARTIK 10.
A better (and yes, more complicated) option is to temporarily store the excess energy and capture it later by using a buck regulator, a type of switching power supply. In a buck regulator, you put a switch and inductor between the battery and load, and a capacitor across the load. When you close the switch, current flows from the battery through the inductor to the load. Increasing current through the inductor induces a magnetic field that stores energy and creates a voltage drop between the battery and load. Current through the inductor powers the load and charges the capacitor.
As the voltage across the load increases, you open the switch. The decreasing current induces a positive voltage across the inductor, drawing current through the diode. For an average load of 700 mW and an average supply of 1100 mW with switch closed, we would operate the buck converter at a 700 / 1100 = 64% duty cycle. The voltage across the load will fluctuate; the magnitude and period of the ripple depends on the frequency at which we open and close the switch, the size of the inductor and capacitor, and the load.
During discharge cycles there will be a voltage drop across the diode, which will decrease the efficiency of the regulator. To boost efficiency, you could replace the diode with a switch and coordinate the timing of the two switches very carefully.
As you can see, designing a buck converter from scratch is complicated. Alternatively, for $5.60 (in thousands) you can buy a TSR 0.5–2433SM buck converter from TRACO power. Add one external trim resistor to set output to 3.4 V and you’re done. The converter requires a minimum of 1.5 V of headroom between input and output, or about 2.5 V per cell of our battery. That’s not as low as the 1.8 V we got with the LDO, but your users don’t care how many electrons we pull from a battery; they only want to know how long a battery will last.
Avoiding user error
Speaking of users, some will accidentally put their batteries in backwards. Best case scenario: their device won’t work until they figure out their mistake, then work perfectly when they turn their batteries around. If you didn’t protect your design against reverse polarity, their device is dead and you will get nasty calls on your help line.
A diode in series with the positive terminal of your battery is the easiest protection against reverse polarity; install the batteries backwards and the diode blocks the current. The catch is that in normal operation you’ll have a voltage drop across the diode of about 0.7 V, which cuts off a bit of the power you can extract from the batteries.
A better alternative is to use a P-channel MOSFET, which has an internal diode that is normally reverse biased between source and drain. If you connect the drain to the battery, source to your regulator circuit, and gate to ground, you forward bias the diode in normal operation. Unlike a standard bi-polar diode, the MOSFET diode offers zero insertion loss (full voltage with no current) and is very low on resistance. Digital DIY has a good explanation of the circuit if you want the details.
What about the sun?
Why even bother with costly, replaceable batteries in your garden weather station when there’s all that free sunshine available? We’ll get into that question in the next tutorial in our IoT 101 series, when we’ll cover alternate power sources for our IoT devices. We’ll cover line power, solar panels, and how to design battery backup systems for when the sun doesn’t shine or the power goes out.
*If you have lessons you’d like to see in this series, ask us @SamsungIoT; and a big shout-out to the developer who suggested this lesson about batteries.
About the author: Kevin Sharp has been an engineer since long before he got his engineering degree, and has extensive experience in data acquisition and control networks in industrial, retail, and supply chain environments. He’s currently a freelance writer based in Tucson, Arizona.t