IoT sensors and hardware: boards, sensors and power

For anything that needs to send data over Wi-Fi, an ESP32 development board is the shortest path: the radio is built in, the boards are inexpensive, and the ecosystem of libraries and examples is large. If the project is about reading sensors and driving simple outputs with no network at all, a classic Arduino board is easier to understand, because there is less going on inside it. The Raspberry Pi Pico sits between the two, with a fast core and flexible input and output but no radio on the base model.

The microcontroller comparison article sets out the trade-offs board by board. The general advice is to choose the board whose examples most closely match what you want to build, because for a first project the quality of the available tutorials matters more than the specification sheet. Buy two of whatever you choose; the second one saves a week when the first meets a mistake with the power supply.

As accurate as the decision you will make from its reading, and no more. A sensor that turns a fan on when a room gets warm does not need to know the temperature to a fraction of a degree, because the fan's threshold is a rough one anyway. A sensor that logs the humidity in a storage room where the acceptable band is narrow needs to be repeatable across that band and to respond quickly when conditions change.

Two things matter more than the headline accuracy figure on a datasheet. Repeatability is whether the same conditions give the same reading each time, which is what an alarm threshold relies on. Response time is how long the sensor takes to follow a change, which decides whether it can catch a short event. The sensor comparison article treats both, and the sensor placement notes in it explain why a well-placed cheap sensor often beats a badly placed accurate one.

Anywhere from a day to several years, and the difference is almost entirely how the node spends its time. A microcontroller with its radio on continuously draws enough current that a small battery lasts hours or days. The same microcontroller that wakes for a fraction of a second, takes a reading, sends it and goes back to a deep sleep state can run for months on the same battery, because the average current is dominated by the sleeping state rather than the active one.

The powering article shows how to estimate this with a simple budget: the active current, how long each wake lasts, how often it happens, and the sleep current in between. It also covers why the battery chemistry and the voltage regulator matter, and when a small solar panel makes more sense than a bigger battery. If you want a single rule, it is that the interval between readings is the lever you can move most.