An analog-to-digital converter chops a continuous signal into a finite number of voltage steps and time samples. The resolution (number of bits) sets how fine the voltage steps are — one step is the LSB — and also the best-case noise floor, because rounding each sample to the nearest step adds quantization noise. The sampling rate sets how fast it captures, and the Nyquist limit (half the rate) is the highest frequency you can faithfully record. These few numbers decide whether a converter is good enough for your signal.
Reviewed: June 19, 2026 · Author: Naveen P N, Founder — AI Calculator · Verified against: Analog Devices / TI data-converter application notes.
The ADC formulas
The 6.02 dB-per-bit relationship comes from each extra bit halving the quantization step, which doubles the signal-to-noise ratio (6.02 dB). The +1.76 dB term is the statistical noise advantage of a sine wave over its quantization error. ENOB simply re-states a measured SINAD in bits, so you can compare a real part against its nominal resolution.
Worked example — choosing an ADC for a sensor
Scenario: A sensor outputs 0–3.3 V and you need to resolve 1 mV with at least 70 dB of dynamic range, sampling at 100 kSPS.
A 12-bit converter gives LSB = 3.3/4096 = 0.806 mV (better than 1 mV) and an ideal SNR of 74 dB — comfortably over the 70 dB target. At 100 kSPS the Nyquist limit is 50 kHz, so an anti-aliasing filter must remove anything above 50 kHz. Remember the real ENOB may be ~10.5 bits, so leave margin.
Frequently Asked Questions
LSB = Vref ÷ 2N. A 12-bit ADC with 3.3 V reference: 3.3 ÷ 4096 ≈ 0.806 mV — the smallest change it resolves.
SNR = 6.02 × N + 1.76 dB for a full-scale sine. 12-bit ≈ 74 dB, 16-bit ≈ 98 dB; each bit adds ~6 dB.
Effective number of bits = (SINAD − 1.76) ÷ 6.02. A 16-bit ADC measuring 90 dB SINAD has ENOB ≈ 14.7 bits.
Half the sampling rate: fNyq = fs ÷ 2. Signals above it alias to false low frequencies unless filtered. 1 MSPS → 500 kHz limit.
≈6.02 dB per bit, so bits = dB ÷ 6.02, rounded up. 90 dB needs ≈15 bits → a 16-bit part. Allow for ENOB being 1–2 bits lower.