Frequency normalized leakage correction technique

Leakage is a problem in nearly all aspects of signal processing, especially when the FFT is involved. Leakage takes the form of wide frequency spikes due to a ‘‘spilling over’’ of information across frequency bins and occurs most often when the sampling frequency is not an integer multiple of the base frequency. The leakage correction method presented here substantially reduces this problem. The method resamples the original time domain data within specified limits. A frequency error measurement is taken for each sampling rate within the defined limits. The minimum frequency error is recorded along with the fundamental frequency of the record. This method resamples the entire signal one record at a time with a linear interpolator. Nyquist criteria must be satisfied at all times to validate the use of the linear interpolation. Nyquist criteria is satisfied if the sampling rate is at least twice as large as the highest frequency in the signal. The resampled data must satisfy three conditions to eliminate leakage. (1) 2n points per period. (2) Integer number of periods per record. (3) Satisfy Nyquist criteria. The method here satisfies these conditions and results in enhanced time domain data. a)Now at Martin Marietta, Knolls Atomic Power Labs, Schenectady, NY. b)Now at Dept. of Mech. Eng., Purdue Univ., West Lafayette, IN.