Predicting Gated-Peak Grain Noise Distributions for Ultrasonic Inspections of Metals

In ultrasonic pulse/echo inspections of metal components, defect detection can be limited by backscattered “grain noise” from the metal microstructure. The absolute level of grain noise observed in a given inspection depends on the metal microstructure and on details of the inspection system, such as the focal properties of the transducer, the spectral content of the incident sonic pulse, and the receiver amplification settings. In earlier work [1–3], we presented models which account for both measurement system and microstructural effects, and which predict certain aspects of the backscattered noise in weakly-scattering materials. For example, one of the predicted quantities is the “rms noise level”, illustrated in Fig. 1 and defined as the root-mean-squared average of the RF noise voltages seen at a fixed observation time when the transducer is scanned above the specimen. The absolute rms noise level as a function of time (or penetration depth) can be predicted from knowledge of the transducer diameter and focal length, a “reference echo” from a flat surface, and certain material properties of the specimen including its density, velocity, attenuation coefficient, and Figure-of-Merit (FOM).