When you build a bunch of transducers, you will get a variability. That’s the real world. The parameter with the most variability is often the Qm, the mechanical quality factor. 

One of the major goals of measurement is to determine tolerances by which we can exclude transducers that are likely to perform poorly.

However, with regards to Qm (mechanical quality factor), we often find lower Qm transducers perform to a similar standard as higher Qm transducers. This is for transducers built with the same methods. 

So, is Qm a bad metric for finding ineffective transducers? Not necessarily. In these cases, it becomes very useful to test the parameter, in this case Qm, over a variety of power levels. Then, instead of looking for a low Qm at a specific power condition to find a bad apple, we observe and document thresholds of abnormal changes.

When measurements at a certain condition can’t be correlated to other behaviors, try varying the measurement conditions (eg power levels) and looking for thresholds.