Calibration drift during a plate-weld scan
During a site inspection of a 25 mm thick carbon steel plate butt weld, you use a 4 MHz, 60° shear-wave probe and a DAC made from 3 mm side-drilled holes. The procedure requires a sensitivity check before and after each 4-hour scan period. After 3 hours, with the probe on the same reference hole at 20 mm sound path, the echo has fallen from 80% to 55% FSH at the same gain setting. The probe delay has shifted by 0.3 µs. You have completed 18 m of weld and recorded two indications that were evaluated as acceptable. A cable connector feels loose. The couplant and surface are unchanged. The client wants to know if the completed length can be accepted and what must be re-examined.
1. What is the significance of the amplitude and delay changes, and what immediate checks do you make?
Read the worked answer
The 80% to 55% drop is about 3.2 dB, which exceeds the normal ±2 dB sensitivity-check tolerance; the 0.3 µs delay shift also changes depth calibration. Stop scanning. Check the loose connector first, then the cable, probe shoe and wear face, instrument battery and couplant. Re-zero the probe on the reference block and repeat the 20 mm SDH check. If amplitude and delay return to the original values after tightening or replacing the cable, the fault is in the connection; if not, substitute a known-good probe and cable. Do not adjust the DAC to hide the change. Only when the reference echo is stable within tolerance over several repeat measurements should scanning resume. Record the fault, the checks and the correction in the inspection report.
2. What is the validity of the 18 m already scanned, and what re-examination is required?
Read the worked answer
Because the last valid check was at the start of the 4-hour period, the drift could have occurred at any time during the 18 m. A 3 dB loss could have caused an indication to fall below the recording level, and the delay shift could have distorted depth sizing. The two indications previously accepted must be re-evaluated with the corrected calibration. Re-scan the whole 18 m at corrected sensitivity; a partial re-scan from the last good check is not defensible unless the cause is proven to be a sudden event after a known point. If the re-scan finds no new recordable indications and the two original indications remain below the acceptance level after re-sizing, the length can be accepted. Otherwise repair or reject and re-examine. Raise a nonconformance and document the re-scan extent.