Every project eventually needs a way to interact, and sooner or later, that means buttons. But wiring buttons can be surprisingly tricky, especially when you want more than a couple. There are several ways to do it, each with its own strengths and limitations, depending on how many inputs you need and how tidy you want your circuit to be.
The simplest method is one button per pin. It’s direct and easy to understand: connect one side to ground, the other to a GPIO with an internal pull-up resistor, and detect when it goes low. It’s reliable, but it scales terribly, fill your project with ten buttons, and you’ve burned ten pins.
Matrix wiring is the next logical step. By arranging buttons in rows and columns, you can drastically reduce the number of pins needed. For example, a 4×4 matrix gives you sixteen buttons using only eight pins. It’s efficient, but it adds some complexity to your code and requires diodes if you want to avoid ghost presses when multiple buttons are pressed at once.
Another common approach uses shift registers or I/O expanders like the 74HC165 or MCP23017. These let you read many buttons using just a few pins over SPI or I²C. It’s flexible and scalable, but adds cost, components, and code overhead, not always ideal for smaller projects.
Then there’s the clever method featured in this Curious Scientist article which uses an analog input to handle multiple buttons through different resistor values. It’s a surprisingly practical trick that keeps wiring minimal while still giving each button its own identity. If you’ve ever run out of GPIOs or just wanted to simplify your design, it’s worth reading about.
Ultimately, the “best” button solution depends on what you’re building. For a small gadget, direct wiring is fine. For something larger, a control panel, a custom keyboard, or a product with strict space constraints, matrix, shift-register, or analog methods can save you serious headaches.
If you’re curious about that last approach and want a detailed breakdown (with real examples and measurements), check out Curious Scientist’s post:
