Blog · · 5 min read
Ten years on one battery: what really limits battery life in a low-power device
A power budget for a battery-powered meter reader holds a surprise: photos and radio use under two percent of the cell. Self-discharge, heat and one small capacitor decide. What AI can calculate, and what you still have to measure.
Every battery-powered gadget runs into the same question: how long will the battery last? The usual answer is to look for the big consumers, the radio, the camera, the lights, and make them efficient.
For ZählerSchau, an open-hardware project on makemachine.ai, that approach would have given the wrong answer. ZählerSchau clips onto an electricity meter, takes a photo of the display on set dates, reads the digits itself and sends the value as one short radio message. The goal is ten years or more on a single AA cell, and the design is checked against 15 years. When the platform worked out the power budget, the camera and the radio turned out to be almost irrelevant. Three other things decide the battery life.
Step one: count states, not parts
A power budget starts with a list of everything the device can be doing, and for how long. For a meter reader that list is short:
- Asleep, almost all of the time. Only a wake-up timer runs.
- Awake for a reading: lights on, photo, the processor reads the digits.
- Sending: one message of about 20 bytes. If unsure about a digit, it adds a tiny grey picture of the digits as proof.
For each state you need a current and a duration. Multiply and add up, with the sleep current around the clock. This is the kind of work AI does well: pulling currents from datasheets, building the table and answering "what if" questions in seconds. What if it reads twelve times a year instead of four? What if the radio needs its slowest setting inside a meter cabinet?
Where the charge actually goes
The budget over 15 years at four readings a year, rounded, for an AA lithium-thionyl chloride cell of roughly 2,500 mAh, with generous numbers for the active parts:
| Where the charge goes (15 years) | Charge | Share of the cell |
|---|---|---|
| 60 readings, each 10 s at 200 mA (lights, camera, processor) | about 33 mAh | about 1.3 % |
| 60 radio messages, each up to 1.8 s at 100 mA | about 3 mAh | about 0.1 % |
| Wake-up timer, always on, about 0.2 µA | about 26 mAh | about 1 % |
| Self-discharge of the cell, roughly 1 to 2 % a year | 375 to 750 mAh | 15 to 30 % |
1.8 seconds is the slowest radio setting; the fastest takes well under a tenth of a second. A proof picture now and then adds a few more messages and barely changes this.
Everything the device actually does adds up to less than two percent of the cell. The cell loses ten times more just by sitting there. That is why the concept for ZählerSchau estimated that one cell would use only about a quarter of its charge in 15 years, and why one or twelve readings a year barely change the result.
Surprise one: the battery dies of old age, and heat speeds it up
Self-discharge grows with temperature. A cool basement is kind to a battery, a sunny utility room or an outdoor box is not. ZählerSchau is designed for 55 °C. At that temperature, the research step estimated that self-discharge alone would take 67 % of the cell over 15 years at typical rates, and 85 % at the maximum rate. An older study even suggests the rate could be much higher for some electrolytes. That is not something to settle with a calculation. The cell maker has to confirm it.
Surprise two: the part that ages before the battery
Lithium-thionyl chloride cells hold a lot of energy and lose little of it, but they are bad at delivering current quickly. A thin layer builds up inside the cell over time, and under a sudden load the voltage dips before it recovers. A burst of light or a radio message is exactly such a load. The usual fix is a pulse capacitor that charges slowly from the cell and delivers the bursts. In the ZählerSchau design, the bare cell is good for about 40 mA continuously, while the capacitor can supply about 2 A for one second and 0.5 A continuously. Those numbers decide the fuse and the board layout.
But the capacitor has its own lifetime. Its stated life is 10 years at room temperature and 7 years at 60 °C, until it is down to 80 % of its capacity. That is shorter than the 15-year design target, and in a hot cabinet it could fall short of even the ten-year goal. The research step flagged it as the biggest new risk of the whole design.
Surprise three: 0.2 µA is a promise the whole board has to keep
The board sleeps at about 0.2 µA, because a nano-power timer switches everything else off between readings. That number only holds if every part on the board keeps it. One forgotten 10 kΩ pull-up resistor held at the wrong level draws 360 µA at 3.6 V. That is 1,800 times the budget, and it would empty the cell in under a year. Other usual suspects: a regulator with a high idle current, a sensor that never sleeps, a status LED, or firmware that forgets to switch something off. None of them is labelled "power consumption" in the schematic. They show up on a current meter.
What AI does well here, and what you still have to do
AI is fast and thorough at the paperwork of a power budget: datasheets, states, arithmetic, what-ifs. In this project it also found the two biggest risks, heat and the capacitor's lifetime, because it checked the sources.
What it cannot do is make a datasheet true. "Typical" values are not maximum values, and many datasheets say little about 55 °C or 15 years. An AI can also misread a table or invent a number, so every figure in a budget should point back to its source (more on that here). Three things stay with people:
- Ask the cell and capacitor makers for data at your hottest temperature and your longest life.
- Measure the current on the first real boards, in every state, with a meter that can resolve fractions of a microamp.
- Test the radio from the real spot. A meter cabinet is often a metal box. If the radio needs its slowest setting there, check the budget again.
For ZählerSchau, these tests on real meters are the next step.
The short version
- Count states and durations, not just parts.
- For a device that wakes up rarely, sleep current and self-discharge decide, not the radio.
- Design for the hottest place the device will live.
- Check the lifetime of every part in the power path, not only the cell.
- Measure the sleep current on the real board before you promise ten years.
You can follow the whole project on the ZählerSchau project page. Planning a battery-powered device of your own? Start with an AI feasibility check, and read what AI can do on the circuit board.
Watch hardware get built by AI. Follow public projects on makemachine.ai, read every result in full and get an email when there is something new.
More from the blog
- Will my electronics enclosure overheat? A five-minute heat estimate for a sealed box
- Joining 3D printed parts: screws, heat-set inserts, snap fits or glue?
- Free or paid AI 3D modeling? What you really pay for, and a 30-minute test to compare tools
Open-hardware projects built with AI
- DIY Automatic Plant Watering System: An open-hardware watering system for indoor plants: a small pump, a refillable tank and a soil moisture sensor in each pot keep up to six…
- DIY BBQ Temperature Controller: A temperature controller for charcoal grills and smokers that makers can build themselves. A pit probe, two meat probes and a small fan on…
- ZählerSchau: ZählerSchau is a small, battery-powered unit that clips onto an electricity meter in Germany (water meters via a clamp adapter), takes a…
- Roadstoy: Roadstoy is a remote-controlled toy version of the new Tesla Roadster that can briefly jump and flip using compressed air. It drives on an…
- The forever Washing machine: The forever washing machine is a washing machine offered as a service instead of a product. Customers do not buy the machine but pay for…
- Bike Light: I want to build a bike light that needs no battery. Magnets sit in the spokes close to the hub, and a coil on the fork picks up their…
- Rooficle: I want to develop a light, efficient and affordable e-bike that can be used legally on public roads in Singapore. Its core is an…
- BBQ Modular Grill: I want to design a modular grill. Instead of one sealed block, it is made of separate parts that can each be bought on their own online…
- Go FANUC Go: A mobile base that moves an existing FANUC M-710iC industrial robot (about 560 kg, with a milling spindle at the wrist) out of a room and…
- Jump & Run Robot: I want to build a small, low-cost toy humanoid robot that people can put together themselves. It should be about 30 cm tall, or as small…