Salutations,
What we're experiencing isn't so much an error as a lack of expected behavior. The big picture is this: we want to generate Bode & Polar plots from proximity probe data recorded during the startup/shutdown of a turbine. The problem is that the data is too much to hold in memory at one time (5kHz sampling over ~15-20 minutes, but potentially much longer). There may be a flaw in the concept I used to create the software. I assumed that if I fed a chunk of data to the resampling vi and then to the order extraction vi I could take the resultant magnitude, phase, and RPM vectors and average them for the chunk of data (ideally 1 second) generating one point for the eventual magnitude and phase XY graphs. In this way, I would build a graph containing 1 point for each second of data that goes in. The assumption is that at a given RPM, the average magnitude and phase should be fairly constant.
Currently, at high RPM values the we seem to be getting answers (are they correct, not entirely certain), however at lower RPM values the data set returns the NaN issue. This could be due to a lack in points being sent into the VI's, as the RPM increases the amount of data information from the tach sensor (over the 1 second) increases. While probing the data after Labview's OAT Resample Waveform, the dr value remains constant during the NaN section, however it becomes dynamic when the data starts to show answers. This all occurs prior to entering the OAT track orders in resample signal vi.
Hence, we are uncertain if there are specific limitations to the vi's that aren't mentioned in the manual. Yes, I RTFM'd. 😉
Thank you for your time and assistance,
E. Smith
