Here are the questions you need to answer: What did the machine do in that time? How many times did the shaft rotate? Could we tell if there were any short-duration impacts? Could we see beating or modulation? We have three parameters under our control: the sample rate (the number of samples per second), the length of the time record, and whether we integrate the signal (when using an accelerometer). If we were to sample the waveform a billion times per second, and do so for one minute, we would have a very clear picture of how the waveform changed as the shaft turned and the balls rolled around the raceway, etc., but our data collector would run out of memory. Or we could sample the waveform at just 20 times per second, for only a tenth of a second – but the waveform would reveal no information at all. Obviously, the ideal setting is somewhere in between. Here we have two time waveforms. This waveform represents 1024 time samples with 0.3 seconds of data. The time waveform does not reveal anything of great interest. It is clear that the machine is not running perfectly smoothly, but it is difficult to learn a lot more from this waveform.
Posted on: Thu, 12 Sep 2013 18:35:54 +0000
Trending Topics
Recently Viewed Topics
© 2015