Microcontroller projects

A microcontroller is a complete small computer on one chip, with pins that connect directly to the physical world. It has no operating system and does one program forever, starting the instant it gets power — which is exactly what you want for a thing that watches a sensor and switches something. That is the whole idea, and everything else is detail.

The distinction worth learning early is between a microcontroller and a single-board computer. A microcontroller runs one program with precise timing and starts in milliseconds; a single-board computer runs an operating system, does many things at once, and is the right answer when you need a network stack, a screen or a filesystem. Choosing wrongly is the commonest architectural mistake in the hobby and it is usually made in the direction of too much computer.

The shape of a project is almost always the same: read something, decide something, change something. A sensor, a rule, an output. Once that loop is working the rest is elaboration — and the honest observation is that most of the learning happens in the first three or four projects, after which the hobby becomes a component of other hobbies rather than a destination.

A capable board costs

£4–£15

The shape of every project

Read, decide, change

The common wrong turn

Too much computer

Where it usually ends up

Inside another hobby

1. 1

   Step 1: Buy a board, a breadboard and a bag of LEDs

   Under £30 all in, from one order. Do not buy a kit containing ninety components you cannot name; buy the four things you will use this week.

   You will know it worked when: You own a board and can power it from a USB port.

2. 2

   Step 2: Blink one LED, and understand every line

   The canonical first program, and the temptation is to skip it because it is trivial. Do not — it exercises the whole chain from editor to chip, and when a later project fails it is the known-good baseline you fall back to.

   You will know it worked when: Something you wrote is running on hardware you are holding.

3. 3

   Step 3: Read a button, and meet debouncing

   A button does not produce one clean press; it produces a burst of contacts as the metal settles. Discovering that, and fixing it, is the moment the hobby stops being magic and becomes engineering.

   You will know it worked when: One press registers once.

4. 4

   Step 4: Close the loop with a sensor

   Read something from the world, decide something, change an output. This is the shape of every project you will ever build, and having it working once means every future project is a variation rather than a mystery.

   You will know it worked when: The device does something without you touching it.

5. 5

   Step 5: Build one thing that leaves the bench

   A device that does a real job somewhere in your house, powered from a real supply and, ideally, in an enclosure. This is the step that separates people who have done tutorials from people who have made something, and it is where all the unglamorous problems live.

   You will know it worked when: Something you built is in use, unattended.

6. 6

   Step 6: Only then learn to solder

   When breadboard wires falling out becomes the limiting factor. With extraction, and on something you do not mind ruining first.

   You will know it worked when: A permanent version of something that already worked.

• A board, chosen for its ecosystem — What matters is not the chip but how much beginner-facing material exists and whether the community answers questions kindly. The two dominant hobbyist families both qualify comfortably and either is a fine first choice; picking either and starting beats a fortnight of comparison, because the concepts transfer almost completely and switching later is cheap.
• A breadboard and jumper leads — Solderless prototyping, which is what makes this hobby forgiving: everything is reversible and nothing is committed. Get more jumper leads than seems reasonable — they are the consumable and they go missing.
• LEDs, resistors and buttons — The unglamorous starter set, and genuinely where the first month goes. An LED that lights when you meant it to is a complete education in output, and a button that reads reliably is a complete education in input, including the debouncing problem nobody warns you about.
• Two sensors, not twelve — Temperature and something that detects motion or distance. The instinct is to buy an assortment pack; the result is a drawer of modules with no projects attached. Buy a sensor when a project needs it.
• A multimeter — The tool that turns guessing into measuring, and the point at which debugging becomes possible rather than mystical. £20–£60 and it will outlast everything else on the bench.
• A soldering iron, later and with extraction — Needed only when a project graduates from breadboard to permanent. Temperature-controlled, because a cheap uncontrolled iron makes intermittently failing joints, which is the worst class of fault to chase. Buy the fume extractor at the same time.
• What can wait indefinitely — An oscilloscope, a bench power supply, a logic analyser, a soldering rework station, and any assortment pack of anything. All are genuinely useful eventually and none of them is what stands between you and a first working project.

And the ordinary things, which are about equipment rather than people:

• Mains is a different hobby — Switching anything that plugs into a wall is electrical work with real consequences, and a relay module does not change that. Use ready-made mains-rated switching hardware, or keep the project on low voltage — which is where essentially everything worth building lives anyway.
• Motors need their own supply — Not a safety point but the commonest way a board is destroyed: a pin switches a signal, not power, and a motor connected directly to one takes the board with it.
• Reverse polarity kills boards silently — Most cheap boards have no protection. A moment checking which way round the supply goes saves the commonest kind of dead board.
• Eyes, when snipping component legs — Trimmed leads travel at surprising speed. The one place safety glasses earn their keep on a bench like this.

• The duty, stated by the regulator — Ofcom puts it plainly in the foreword to the document that governs this: "It is required by the Wireless Telegraphy Act 2006 that no radio equipment is installed or used in the UK except under the authority of a licence granted by or otherwise exempted by regulations made by Ofcom." Almost everything a hobbyist uses is exempt rather than licensed — but exemption is conditional, and the condition is that the equipment "must meet the minimum requirements specified in this UK Interface Requirement for the stated equipment types and for the stated frequency bands".
• Why a CE or UKCA mark on the module is not the whole answer — The mark says the product may lawfully be placed on the market. The interface requirement is about how it is used, in which band and at what power — and a module that is compliant as sold can be configured, in software, to operate outside the exemption. On a development board, changing the band and the transmit power is two lines of code.
• The four bands hobby projects actually land in — 2400–2483.5 MHz is where Wi-Fi and Bluetooth sit, at 100 mW for wideband data, or 10 mW for non-specific use. 868–869.7 MHz is the European short-range band at 25 mW e.r.p. 433.05 MHz upwards carries the cheap remote-control and sensor modules at 10 mW e.r.p. with a duty cycle limit. And 915–918 MHz is exempt at 25 mW e.r.p., again with a duty cycle limit.
• The 915 MHz claim, corrected — It is repeated constantly in maker communities that the 915 MHz LoRa modules sold for North America are simply illegal in Britain. That is not what the document says, and the true position is more useful. The UK exemption does cover 915–918 MHz — but at 25 mW with a duty cycle limit, where the North American profile a module ships on typically spans 902–928 MHz at considerably higher power. The band overlaps; the conditions do not. A module left on its factory US setting will be operating outside the exemption at both edges of the band and above its power limit, which is a different and more precise problem than "the frequency is banned".
• What is not covered here at all — Amateur radio, which is a licensed activity with its own examinations and privileges and is not what this page is about. Anything you intend to sell, which brings the Radio Equipment Regulations and conformity assessment into it. And antenna design, transmit power measurement and how to establish that a given build actually complies — which are engineering questions this page has no basis to answer.

• Data logging and environmental sensing — The natural first real project: measure something over weeks and look at the numbers. Cheap, genuinely useful, and it teaches storage, timing and power in one go.
• Custom input devices — Keypads, macro pads, MIDI controllers, game controllers. A board that presents itself as a USB device opens a large and satisfying category.
• Wearables and e-textiles — Conductive thread, sewable boards and small batteries. A genuinely different discipline that meets textiles rather than electronics, and one where the constraints do the design.
• Retro and emulation projects — Reproducing or interfacing with old hardware, which is where a microcontroller's precise timing matters in a way a full computer cannot easily match.
• Robotics, which owns the moving half — Once a project drives motors and senses its way around, it is robotics and that page covers it properly — drivetrains, control loops, and the branches from line followers to combat machines. This page stops where that one starts.
• Home automation, which owns the house — A board reporting to a hub is home automation, and that page covers the hub decision, the local-versus-cloud question and the electrical boundary. Again, this page does not restate it.

This hobby exists in its current form because of a deliberate cultural choice made by other people: open-source hardware designs, permissively licensed libraries, published schematics, and a vast body of tutorials written by hobbyists with nothing to sell. Nearly everything a beginner touches — the board, the library that drives the sensor, the wiring diagram — was given away. The norms below follow from that.

• Credit and licence, even for a shed project — If you publish a project built on somebody’s library, you say whose it is and you respect the licence. It costs a line in a README and it is the entire mechanism by which the commons keeps working. The community is relaxed about almost everything and firm about this.
• Ask with a schematic, a photograph and what you already tried — The single most useful convention to learn. "It does not work" gets nothing; a wiring photograph, the code, the error, and the two things you have already eliminated gets a detailed answer within the hour. This reads as gatekeeping and is not — it is that electronics faults are almost impossible to diagnose from a description.
• Answer the beginner question you were asked last year — The reciprocity norm, and it is genuinely observed. Somebody two months ahead of you is often the best person to answer your question, and answering is how people move from consuming the commons to maintaining it.
• Clones are legitimate, and the etiquette is to buy an original occasionally — Open hardware means compatible boards are legal and cheap, and nobody is judged for using them — the licences permit exactly this. The convention, stated often and enforced by nobody, is that if a project ends up mattering to you, you buy something from the people who designed it.
• Publish the failure as well as the build — A strong norm in the maker community and an unusually valuable one. Write-ups that include the version that caught fire, the pin you shorted and the three weeks lost to a floating input save more people’s evenings than the clean final photograph does.
• Makerspaces have their own layer of this — If your route in is a hackspace rather than a kitchen table, expect induction before you touch anything with a motor, a culture of leaving tools better than you found them, and an expectation that you will eventually teach an induction yourself. Most run on volunteer labour and very thin margins, and they are frank about it.

The first stage is the loop: read, decide, change. It takes a handful of evenings and it is most of the conceptual content of the hobby — everything after it is a wider vocabulary rather than a new idea.

The second is making something permanent and leaving it running. Soldering, a supply that is not a laptop, an enclosure, and the discovery that a project which works on a bench for ten minutes is a different thing from one that works in a cupboard for six months. Power, reset behaviour and what happens after a crash all become real here, and this is the stage that separates the hobby from its tutorials.

After that it stops being a destination and becomes an ingredient. Most people who stay with it end up using boards inside something else — a telescope mount, a fermentation cabinet, a model railway, an instrument — and the depth they acquire is in that other subject. That is a completely legitimate end state and is worth saying, because the hobby is unusual in being most valuable when it disappears into something you already care about.