Frequency domain methods for reducing transducer variability

A major objective of quantitative nondestructive evaluation is to formulate defect detection, classification, and sizing algorithms that are insensitive to variations in transducer characteristics, material type, and defect depth. With the data used in this research, ultrasonic signals were found to vary significantly with changes in the transducer and only secondarily with changes in material and depth. It is shown that the method for minimizing signal variations due to transducer and material changes is to deconvolve the test signal with respect to the transducer response from a reference defect in a block of the same material. Since depth variation primarily affects signal amplitude and has minor impact on shape, detection, classification, and sizing, insensitivity to depth is achieved by avoiding the use of amplitude-dependent parameters. The notion of a 'standard transducer' is introduced. Its mathematical properties and methods of realization are given. It is shown how the effects of variability from different test transducers can be removed by signal processing. When these procedures are applied to the test transducer, the effect is to cause it to resemble the standard transducer and, thereby, to place all ultrasonic waveforms on a common basis.