SPI vs I2C: When to Use Each Protocol

#Two Buses, Two Philosophies

SPI and I2C are the dominant serial protocols for connecting microcontrollers to sensors, displays, and memory. Both are widely supported, but they trade off speed, complexity, and pin count differently. Choosing the right one depends on your constraints.

#SPI: Speed and Simplicity

SPI is fast — often 10–50 Mbps or higher. It uses four signals: clock (SCK), master-out-slave-in (MOSI), master-in-slave-out (MISO), and chip select (CS) per slave. No pull-ups, no addressing. Each transaction is point-to-point. SPI excels for high-bandwidth peripherals: flash memory, high-resolution ADCs, displays.

The downside: pin count scales with slave count. Six SPI devices need six chip selects. On pin-constrained MCUs, that adds up. SPI also has no formal standard — different vendors use different polarities and phases. Always check the datasheet.

#I2C: Fewer Wires, More Overhead

I2C uses two wires (SDA, SCL) and addresses each device with a 7- or 10-bit ID. Up to 127 devices on one bus. Great when you have many low-speed sensors: accelerometers, temperature sensors, RTCs. Pull-ups on both lines are required; typical values are 4.7kΩ.

I2C is slower — 100 kHz standard, 400 kHz fast mode, 1 MHz high-speed. There's protocol overhead: address, acknowledgments, repeated starts. For a single high-speed device, SPI is usually better. For a bus full of sensors, I2C keeps your schematic clean.