Engineering
How a Sigma-Delta Analog-to-Digital Converter Achieves High Resolution
Quick fact
A sigma-delta ADC can achieve 24-bit resolution using only a 1-bit quantizer (a comparator) by oversampling and shaping the quantization noise so that most of it falls outside the signal band, leaving an almost noise-free signal.
Why this is interesting
You might think that to get a 24-bit digital reading of an analog signal you need a circuit that makes 16 million distinct decisions. But the ADC in your smartphone's microphone achieves that precision using a single bit and a lot of clever repetition—how?
Read the full explanation
Understanding How a Sigma-Delta Analog-to-Digital Converter Achieves High Resolution
Imagine you are trying to measure the exact height of water in a glass by taking a photo every second. If you only have a black-and-white camera with one pixel that says 'above the line' or 'below the line', you can't tell the exact height from one snapshot. But if you take many snapshots while gently rocking the glass, the proportion of 'above' responses indirectly reveals a very precise average level. That is the core idea of a sigma-delta ADC: it uses a simple 1-bit comparator (the one-pixel camera) but samples the signal many times faster than the actual rate you need. The analog input passes through an integrator, which accumulates the difference between the input and a feedback signal derived from the comparator output. The comparator decides whether the accumulated value is above or below a threshold, outputting a 1-bit stream. This feedback loop continuously steers the integrator to track the average input. By averaging many 1-bit decisions in a digital filter, you recover a high-resolution digital value. The magic is that the averaging not only smooths the input but also dramatically reduces the noise introduced by the coarse 1-bit quantization, because that noise gets pushed to very high frequencies and is then filtered out.
A deeper explanation
The secret behind sigma-delta's high resolution is a technique called noise shaping. In a basic quantizer, the error (the difference between the actual analog value and the quantized digital value) is uniformly spread across the frequency spectrum. A sigma-delta modulator uses feedback to change that. The integrator acts as a low-pass filter for the signal but a high-pass filter for the quantization noise. Mathematically, the quantization noise is pushed to higher frequencies, away from the signal band of interest. Because the ADC samples at many times the Nyquist rate (oversampling), the signal band occupies only a small fraction of the total frequency spectrum. After the modulator, a digital low-pass filter (the decimation filter) removes the high-frequency noise and reduces the sample rate back down to the Nyquist rate. The remaining in-band noise is dramatically smaller than it would be in a non-oversampling ADC, giving the sigma-delta its high effective number of bits. The order of the integrator (first, second, or higher) improves the noise shaping further, but always with a trade-off in stability and complexity. This clever combination of oversampling, feedback, and digital filtering allows a simple 1-bit comparator to achieve resolutions that would otherwise require an impractical number of analog components.