dBFS
How dBFS Anchors Digital Signal Level to the Converter Ceiling
Unlike absolute units such as dBm or dBW, dBFS has no physical power reference of its own. Its zero point is defined entirely by the data converter: 0 dBFS is the largest amplitude the device can represent before its output codes saturate. For an N-bit ADC operating from negative full scale to positive full scale, that ceiling corresponds to the full code range, and every captured sample is expressed as a ratio to that maximum. The result is a scale where 0 dBFS is the absolute top and all usable signal levels are negative numbers. This makes dBFS the natural language for describing levels inside a digitizer, software-defined radio, or DAC transmit chain, where the immediate concern is how close the waveform sits to the clipping point.
The dimensionless nature of dBFS is both its strength and its trap. It lets you reason about headroom and clipping without caring about the analog front-end gain, but it tells you nothing about the actual power at the antenna until you tie 0 dBFS to a physical reference. RF digitizer datasheets close that gap by specifying a full-scale input power, commonly in the +1 to +10 dBm range for a 50 Ω input. Once that anchor is known, a dBFS reading converts directly to dBm by adding the full-scale dBm value, and any analog gain ahead of the converter is subtracted to refer the level back toward the source.
Crest factor is the other concept that makes dBFS practical. A pure sine wave has a 3.01 dB crest factor, so a tone that peaks at 0 dBFS measures -3.01 dBFS RMS. Modulated waveforms have far higher crest factors: an OFDM signal can reach 10 to 13 dB. To keep those rare peaks from hitting the 0 dBFS limit, the average level must be backed off by the crest factor, which is why a high-throughput receiver may run its wanted signal near -15 dBFS RMS while reserving the top of the range for peaks and strong interferers.
Converting dBFS to Absolute Power
The conversion only requires the converter full-scale power and the analog gain preceding it. If a digitizer is rated +4 dBm full scale and a captured carrier reads -18 dBFS, the level at the ADC input is +4 - 18 = -14 dBm. Subtract 30 dB of low-noise gain ahead of the ADC and the signal at the receiver input is -44 dBm. This chain is how dBFS readings feed link budgets and dynamic-range planning.
dBFS in Headroom and Dynamic-Range Planning
Operating point selection in a digital receiver is a balance between two penalties. Drive the signal too close to 0 dBFS and transient peaks clip, generating harmonics and spurious that wreck spurious-free dynamic range. Back off too far and the signal sinks toward the quantization and thermal noise floor, throwing away effective bits. Engineers typically target a wanted-signal RMS level several dB below the crest-factor backoff so that both clipping and noise-floor degradation stay within the link budget.
dBFS Reference Equations
LdBFS = 20 × log10(Vsignal / VFS) dB
Full-scale sine, peak vs RMS:
Peak = 0 dBFS → RMS = 20 × log10(1 / √2) ≈ −3.01 dBFS
dBFS to absolute power:
PdBm = LdBFS + PFS,dBm
Required RMS backoff for a waveform:
Backoff ≈ Crest Factor (dB) → LRMS ≤ −CF dBFS
Where VFS = full-scale amplitude, PFS,dBm = converter full-scale power, CF = peak-to-RMS crest factor. Example: +4 dBm full scale, reading −18 dBFS → P = −14 dBm at the ADC input.
dBFS Compared With Other Decibel Units
| Unit | Reference (0 point) | Absolute or relative | Typical domain | Sign of real values |
|---|---|---|---|---|
| dBFS | Converter full-scale code | Relative to full scale | ADC / DAC / SDR digital level | Negative (0 = clip) |
| dBm | 1 mW | Absolute power | RF power, link budgets | Positive or negative |
| dBW | 1 W | Absolute power | Transmit power, radar | Positive or negative |
| dBc | Carrier power | Relative to carrier | Spurious, phase noise | Negative |
| dBV / dBu | 1 Vrms / 0.775 Vrms | Absolute voltage | Analog audio / instrumentation | Positive or negative |
Frequently Asked Questions
How do I convert dBFS to dBm at a receiver input?
You need the digitizer full-scale input power, often near +4 dBm into 50 Ω (about 1 Vpp full scale). Once 0 dBFS is anchored to that figure, PdBm = dBFS + PFS,dBm. A tone at −20 dBFS therefore sits at +4 − 20 = −16 dBm at the ADC. Subtract any analog gain ahead of the converter to refer the level back toward the antenna, and always take the full-scale dBm value from the datasheet.
Why are dBFS values always negative for real signals?
0 dBFS is defined as the largest amplitude the converter can encode without clipping. Any real signal sits below that ceiling, and the logarithm of a fraction of full scale is negative. A sine touching the limits reads 0 dBFS peak but −3.01 dBFS RMS, since a full-scale sine has an RMS of 1/√2 of its peak. Practical RF digitizers run with several dB of headroom, so typical levels fall in the −10 to −30 dBFS range.
What is the difference between peak dBFS and RMS dBFS?
Peak dBFS compares the instantaneous sample to full scale, so a full-scale sine reads 0 dBFS peak. RMS dBFS compares the root-mean-square level, which for that sine is −3.01 dBFS because of its 3.01 dB crest factor. For high-crest waveforms such as OFDM (10 to 13 dB crest factor), the average must be backed off to roughly −10 to −13 dBFS RMS to keep peaks from clipping at 0 dBFS.
What happens to a signal that exceeds 0 dBFS?
The converter saturates: samples above full scale are hard-clipped to the maximum code, which flattens the waveform peaks and injects broadband harmonic and intermodulation products. In an RF receiver this collapses spurious-free dynamic range and can desensitize adjacent channels. Even brief overranges matter, so digitizers include overrange flags and engineers reserve crest-factor headroom below 0 dBFS.