On the crest factor of noise in full-scale supersonic jet engine measurements

An important consideration in characterizing noise from heated, supersonic jets is the crest factor (CF). The large CF in high-speed jet noise is the result of a positively skewed probability density function for the waveform, which translates into infrequently occurring, large-amplitude positive peak pressures. Sufficient system headroom is required in the data acquisition system to provide an accurate representation of these peak pressures and thus avoid clipping or microphone saturation/distortion. But the question remains as to the importance of capturing the single largest pressure out of potentially millions of waveform samples or if a percentile-based CF is adequate. Measurements near a static tactical aircraft reveal CF increases with engine power, with the maximum CF directed upstream of the overall sound pressure level, and a maximum CF of 20 dB at full afterburner. Second, clipping of measured waveforms at different thresholds reveals that a CF definition based on the 99.99 percentile is sufficient to represent overall and band pressure levels to within 0.1 dB and waveform and time-derivative skewnesses to within ~1%. If an estimate of the time-derivative kurtosis is needed within 1% accuracy, then the 99.999 percentile CF is required for headroom estimates.