Signal Amplitude Calibration Method

It answers a question inspectors ask me constantly. If a weld reflects 20 percent of the signal, how does the wave get past nineteen of them and still find a defect at the far end?

The answer is that a 20 percent reflection coefficient does not mean 20 percent of the wave is gone.

The reflection coefficient is an amplitude ratio, and it is the correct quantity to report. But the energy reflected is that coefficient squared. So a weld with r = 20 percent reflects 0.2 squared, which is 0.04 — four percent of the energy. Ninety-six percent of the energy continues past it, and in amplitude terms 98 percent continues.

Carry that through two welds. In energy: 100 percent arrives, 4 percent reflects, 96 percent continues; at the second weld 4 percent of that reflects, leaving 92.2 percent. In amplitude: 98 percent times 98 percent, which is 96 percent still travelling.

Now carry it through nineteen. The amplitude transmitted is 0.98 to the nineteenth power — 68 percent still travelling after nineteen welds.

Twenty percent reflected sounds like a large loss. Four percent of the energy is not.

That is why welds are not what limits your inspection range. And it is why each weld is useful rather than a nuisance: it gives you a reference signal, and it gives you evidence that the wave reached that far.

The module also covers the three methods of amplitude calibration, in order of how much you have to assume. Calibrating on a weld means accepting a reference that varies from 10 to 25 percent depending on the cap profile. Calibrating on a hose clamp you attached yourself is better, because it is an object of known dimensions. Direct calibration is best: measure the transmitted signal with a plate probe in pitch-catch mode, and you are measuring the wave the probe actually put into the pipe rather than inferring it from something that reflected part of it.

A reference you brought with you beats a reference you found on the pipe. A measurement of the transmitted wave beats both. And whichever you used, say so in the report — every percentage in it inherits the uncertainty of that reference.