Use the criteria for the thinner plate for most applications. The thinner member is the "weak link". The size of the acceptable discontinuity is more stringent for the thinner member when compared to the acceptance criteria of the thicker member if the difference is great enough to place it into a different acceptance category.

It is difficult to give a specific answer if no code is referenced. However, generally speaking the acceptance criteria is based on the amplitude of the signal produced by a reflector (discontinuity). The larger the discontinuity, the higher the amplitude of the reflected signal. That being the premise, a reflective surface of 1/16 square inch will produce a signal with an amplitude of let's say 30% of the full display. A reflective surface of 1/8 square inch will double the amplitude (an increase of 6 dB)of the reflected signal. So, a rating of +3 may be the limit for 1/2 inch thick material. Whereas the acceptable rating for the 1/4 inch thick material may be +9 (increase of 6 db) indicating the acceptable size of the reflector is 1/2 of that for the 1/2 inch thick material.

Best regards - Al

Many thanks for your answer, I was testing some horizontal welds of plates of different thickness, and receive a high peak in the toe of the weld, I check it with straight beam, and get nothing, and then I check a portion of the weld that has been radiographiated, and got the same high peak, I analyze in both size of the weld getting a high peak in the opposites sides of the weld, I just figure that it would be an indication because the change of thickness, but I may be wrong.

When your reflector is producing the highest amplitude, tap the area of interest with your finger tip. If you are in the correct location, i.e., where the sound is reflecting off the surface, the amplitude will be reduced. Your finger tip will dampen the signal so you have a better idea of what is causing the signal. You can then decide whether it is due to geometry or whether there is a discontinuity. Generally, if the signal is fairly uniform for the entire length and if you can dampen the signal, you have a reasonable argument that the signal is due to geometry.

Best regards - Al

again thank you very much