The encyclopedia · Engineering & Operations · Technical decision · 1982-2006
Philips' I2C won chip-to-chip wiring with just two wires and an address.
Philips' 1982 I2C bus ran many peripherals on one clock plus one data line, each chip addressed by the protocol, then made it license-free.
Philips Semiconductors · NXP Semiconductors
the move
Philips Semiconductors (now NXP) invented I2C in 1982 as a simple internal bus to connect low-speed peripherals to a main chip without dedicating a pin and wire to each one.
The protocol needs only an SDA data line and an SCL clock line between devices. After an address byte selects a device, data is clocked in or out; the 8th bit of the address byte flags a read or write, and 7 bits give up to 128 addresses.
Because it used so few wires and no expensive parts, I2C stayed in continuous use, and since October 6, 2006 it has been free to implement without a license. Subsets like TWI and SMBus, and variants up to 10-bit addressing and faster clocks, kept it the default low-speed interconnect.
why it works
- Two wires connect a dozen chips, saving pins on a constrained board
- Addressing is in the protocol, so devices need no separate select lines
- It is slow but free of the four-wire cost of SPI and the one-to-one limit of RS-232
- Philips published it and later removed the licensing fee
what transfers
When you can't add pins or wires, design the protocol to carry addressing itself; a tiny, cheap bus can outlast faster rivals that cost more to wire.
what came after
I2C became the standard way to attach sensors, displays, EEPROMs and management chips to microcontrollers, and 40-plus years of standardization made devices largely interoperable. It is ubiquitous in embedded systems, phones and automotive electronics.
references
spotted an error? The archive wants to know.