Analog-to-Digital Converter (ADC)

How an ADC works

An analog-to-digital converter turns a changing physical signal into discrete digital values that software can process. It does this by taking samples at regular intervals, then assigning each sample a numeric level based on the converter’s resolution and reference range.

This basic pipeline, sample, hold, quantise, encode, is what lets controllers, processors and telemetry systems work with real-world inputs such as sound, voltage, current, pressure, light or sensor feedback.

Sampling, quantisation and resolution

The most important technical trade-off in an ADC is that it cannot preserve an infinitely smooth signal. Sampling decides when the signal is measured, while quantisation decides how precisely each measurement is represented. Higher bit depth gives finer granularity, but it also increases the size of the digital output and can raise implementation complexity.

In practice, two ADCs with the same input may produce very different results if their sampling rates, resolution, input range or noise characteristics differ. A design that is adequate for temperature monitoring may be poor for fast-changing waveforms or precision instrumentation.

Where ADCs appear in systems

ADCs sit at the boundary between the analogue world and digital systems. They are common in embedded controllers, industrial instrumentation, medical devices, communications equipment, audio interfaces and data acquisition hardware. In each case, the converter determines how faithfully the digital system can observe the outside world.

Because the output becomes the input to downstream processing, ADC behaviour affects more than raw measurement quality. It influences filtering, calibration, control loops, anomaly detection and any software logic that relies on sensed values. A weak or poorly chosen converter can create misleading data even when the rest of the system is well designed.

Why ADC choice matters

ADC selection is not just about “making analogue data digital.” It is about matching fidelity, speed, power use, noise tolerance and cost to the task. A converter that is too slow can miss important signal changes, while one with insufficient resolution can flatten small but meaningful variations.

Engineers also have to account for aliasing, reference stability, input conditioning and signal noise. Those factors can make a technically correct conversion still produce a practically unreliable result, especially when the measured signal is weak, rapidly varying or electrically noisy.

Related resources from NHI Mgmt Group

• Guide: AD CS Security Guide
• Why do digital supply chains create higher cyber risk than traditional analog supply chains?
• What is the difference between identity forensics and standard digital forensics?
• How should organisations govern access across many APIs in a digital transformation programme?
• Why does digital transformation make identity governance harder?