Analog I/O (ADC/DAC)

UART, SPI, and I2C move discrete digital bytes. But the physical world — temperature, light, sound, voltage from a sensor — is continuous. Analog-to-digital converters (ADCs) and digital-to-analog converters (DACs) are the bridge between the two.

How an ADC works

An ADC samples a continuous voltage at discrete points in time, then quantizes each sample into one of a fixed number of digital levels determined by its resolution in bits.

voltage resolution = V_ref / 2^n

Worked example: a 12-bit ADC with a 3.3 V reference has:

voltage resolution = 3.3 / 4096 ≈ 0.8 mV per step

Any real-world voltage gets rounded to the nearest 0.8 mV step — this rounding error is called quantization error, and it's an inherent limit of digitizing an analog signal, not a flaw in any particular ADC.

Most microcontroller ADCs use a successive approximation (SAR) architecture: it compares the input against a reference voltage one bit at a time (like a binary search), converging on the closest digital value in a fixed number of clock cycles.

Sampling rate matters too

Resolution determines how precisely each sample is measured; sampling rate determines how often you measure. To reconstruct a signal without ambiguity, you must sample at more than twice its highest frequency component (the Nyquist rate) — sampling slower than that aliases higher frequencies down into your measurement as false low-frequency noise.

DACs: the reverse process

A DAC takes a digital value and produces a corresponding analog voltage, typically by weighting a reference voltage across a resistor network or switched capacitor array. The main practical spec besides resolution is settling time — how long the output takes to stabilize after the input code changes, which matters when generating fast-changing waveforms.

Practical considerations

• Reference voltage stability sets the ceiling on accuracy — a noisy or drifting V_ref corrupts every conversion regardless of the ADC's nominal resolution.
• Input impedance of the ADC's sampling capacitor interacts with the source impedance of whatever you're measuring; a high-impedance sensor may need a buffer amplifier to drive the ADC accurately.
• Oversampling — taking many more samples than strictly needed and averaging them — trades sample rate for effective resolution, a common way to squeeze extra precision out of a cheap ADC.

Why this matters in practice

Nearly every sensor interface in embedded systems — a thermistor, a microphone, a current-sense amplifier — ultimately comes down to correctly configuring and reading an ADC channel, and understanding why the reading has the noise and resolution limits it does.

With digital I/O, serial communication, and analog conversion covered, the last building block is timing: the next sub-lesson, Timers & Interrupts, covers how firmware reacts to events precisely and on schedule.