What a microcontroller is

An embedded system is a computer built into a product to do one job: a washing machine controller, a bike light, a thermostat, a drone. It has no screen or keyboard of its own, it starts the moment power arrives, and it runs the same program for years.

The computer inside is usually a microcontroller, often written MCU. One chip holds the processor core, the memory and the peripherals (the parts that talk to the outside world: pins, timers, serial ports, analog inputs). Popular families for learning are the ARM Cortex-M chips (STM32, nRF52, RP2040), the AVR chips in classic Arduino boards, and the ESP32.

Two kinds of memory matter from day one. Flash keeps your program and your constants, and it keeps them when the power is off. RAM holds the variables your program changes while it runs, and it holds nothing useful at power up until the startup code fills it in. Flash is bigger, RAM is faster, and both are small: a beginner board may have 256 KB of flash and 32 KB of RAM, hundreds of thousands of times less than a laptop.

On Cortex-M chips the big regions are fixed by ARM, and you will meet them in every debugger: the low addresses are for code, RAM starts at 0x20000000, and the peripherals start at 0x40000000. Where exactly the flash sits inside the code region is up to the chip maker: 0x08000000 on STM32, 0x10000000 on the RP2040, 0x00000000 on nRF52. A pointer value tells you what kind of memory it points at just from its first digits.

When the chip starts, a little startup code copies the initial values of your initialised global variables from flash into RAM, sets the globals that have no initial value to zero, and then calls main(). Everything else in RAM is left as it is. Local variables of a function live on the stack, a region of RAM that grows and shrinks as functions are called and return. Then your main() runs a loop that never ends. There is no operating system to return to.