Free · No login required
ADC noise budget
Compare ideal quantization noise with an analog noise floor and estimate the resulting signal-to-noise ratio.
Bit depth (integer): 2–32; Signal level (dBFS): -120–0; Analog noise floor (dBFS): -200–-20
Inputs stay on this page and clear on refresh.
ADC noise budget
Bit depth (integer): 16 Signal level (dBFS): -18 Analog noise floor (dBFS): -100
- Combined noise floor (dBFS)
- -95.930846
- Estimated signal-to-noise ratio (dB)
- 77.930846
Ideal quantization noise Q = −[20 × bits × log10(2) + 10 × log10(1.5)], approximately −(6.02 × bits + 1.76) dBFS. Combined floor = 10 log10(10^(Q/10) + 10^(analog floor/10)); SNR = signal level − combined floor. All levels use a full-scale sine RMS reference of 0 dBFS and the same unweighted, full Nyquist bandwidth; the analog floor must exclude quantization noise to avoid double counting. Enter integer depth 2–32 bits, signal −120–0 dBFS and analog floor −200–−20 dBFS. The model assumes uniform quantization error and independent noise powers. It estimates a noise budget, not actual converter performance, ENOB, distortion, perceptual quality or a recommended recording level. Low-level periodic signals can violate the uncorrelated-error assumption; negative SNR simply means signal below the estimated floor. Dither, noise shaping, weighting, filtering/process gain, jitter and clipping are not modeled. Verify real equipment with measurements. Display uses six decimals without intermediate rounding.
Methodology
How it works
Ideal quantization noise Q = −[20 × bits × log10(2) + 10 × log10(1.5)], approximately −(6.02 × bits + 1.76) dBFS. Combined floor = 10 log10(10^(Q/10) + 10^(analog floor/10)); SNR = signal level − combined floor. All levels use a full-scale sine RMS reference of 0 dBFS and the same unweighted, full Nyquist bandwidth; the analog floor must exclude quantization noise to avoid double counting. Enter integer depth 2–32 bits, signal −120–0 dBFS and analog floor −200–−20 dBFS. The model assumes uniform quantization error and independent noise powers. It estimates a noise budget, not actual converter performance, ENOB, distortion, perceptual quality or a recommended recording level. Low-level periodic signals can violate the uncorrelated-error assumption; negative SNR simply means signal below the estimated floor. Dither, noise shaping, weighting, filtering/process gain, jitter and clipping are not modeled. Verify real equipment with measurements. Display uses six decimals without intermediate rounding.
Method and assumptions documented by Techrider.live
2026-11-02
Frequently asked questions
Does increasing bit depth always improve this result?
No. Once analog noise dominates, reducing ideal quantization noise barely changes the combined floor. The entered analog noise must be measured or estimated separately with matching bandwidth and reference.