Also known as: noise blanking, impulse blanker, pulse blanker
A noise blanker detects brief, high-amplitude impulse-noise spikes and momentarily gates them out of the receive path so they never reach the demodulator, removing the sharp clicks that impulsive interference would otherwise produce.1 It works in the time domain and targets noise that is short compared with a symbol — ignition sparks, power-line arcing, electric fences, and switching supplies — rather than the steady hiss of the noise floor. Because an impulse concentrates its energy in time, it is easier to reject where it is concentrated: with a fast switch that blanks the offending sample.
How it works
A noise blanker has two parts: a detector and a gate. The detector forms a fast estimate of the signal envelope and compares it to a threshold set a few dB above the running average level. Because a genuine impulse rises far above the ordinary signal-plus-noise envelope, it crosses the threshold cleanly. When it does, the gate blanks a short window — typically it holds the sample at zero (a hard blank) or, better, replaces the interval with the surrounding average so the sudden zero does not itself create a spectral splatter. The window is deliberately narrow, just long enough to cover the pulse plus its filter ringing.
The subtlety is timing. By the time the envelope detector fires, the impulse has already entered any narrow filter ahead of it, and a filter smears a sharp pulse into a longer ring. Effective blankers therefore tap the signal at a wide bandwidth — ahead of the channel filter, where the impulse is still short — so the blank window can be brief. This is why a blanker built into a wideband SDR front end can outperform one buried after a narrow IF filter.
The limits are important. A blanker helps only when noise is impulsive and sparse; it does nothing for continuous interference or for raising the noise floor. Set too sensitively, it blanks on strong wanted signals or on modulation peaks, punching holes that degrade reception and can create intermodulation-like artefacts from the switching itself. It is a scalpel for clicks, not a general noise reducer.
Variants
- Time-domain blanker — the classic form described above: an envelope detector and a fast gate operating on the raw sample stream. Cheap and effective against sharp, isolated pulses.
- Wideband (“noise-derived”) blanker — takes its detection cue from a separate wide-bandwidth path so it can catch the impulse before the narrow channel filter smears it, then blanks the main path. This is the arrangement that copes with the filter-ringing problem.
- Frequency-domain / spectral blanking — in an FFT-based receiver, an impulse spreads across all bins of the block it lands in, so a detector can flag and zero the affected transform blocks or use interpolation across them. This suits SDRs that already run block transforms and can target periodic noise (like power-line hash at twice the mains frequency) by its timing.
Blankers are frequently paired with, but distinct from, a noise reducer (an averaging or spectral-subtraction stage that lowers steady hiss): the blanker removes short spikes, the reducer smooths continuous noise, and the two solve different problems.
Relevance to SDR
Noise blankers are standard in HF and VHF communications receivers and are a common feature in SDR software (SDR#, SDRangel, GQRX, and ham transceivers), where mobile installations plagued by ignition noise benefit most. In digital land-mobile decoding the payoff is indirect: an un-blanked impulse can flip symbols and break frame sync, so removing it lowers the bit-error rate on marginal signals. GopherTrunk does not implement a dedicated noise blanker in its DSP chain — its focus is channelization and digital demodulation — so impulse mitigation, where needed, is best handled upstream at the SDR/front-end stage before the I/Q reaches the decoder.
Sources
-
Noise blanker — Wikipedia, on time-domain gating of impulse noise ahead of demodulation. ↩