While I use envelope acceleration, and believe it is a useful means of identifying rolling element bearing degradation... I don't believe amplitude thresholds can be meaningfully applied. The raw signal collected is at a relatively high frequency. I believe, for example, the SKF 3rd band is from 500kH to 10,000kH. These frequencies are easily attenuated... by distance, jointed interfaces, and transducer mounting. Given variables in the amount of signal attenuation... I believe it is impossible to measure severity of degradation by amplitude alone.