LabVIEW
Link FFT with ADXL345 error "you have connected two terminals of different types"
wiebe@CARYA wrote:
I don't have an alternative either. I'd assume you would want to find the direction with the largest vector, and that probably isn't X, Y or Z.
How about the 3D "magnitude:, i.e. space diagonal.
wiebe@CARYA wrote:
@Louay02 wrote:https://www.youtube.com/watch?v=AYvhTv3bsPI
First, that isn't how you should program LabVIEW (sequences and locals).
Second, notice at 1m22, how the cluster of x, y, z, is 'reduced' to either x, y or z.
Horrible code, bad coding habits, horrible music, no narration, no explanation. No links to example code.
I would probably put this video on my personal "Do not watch" list. 😄
@Louay02 wrote:The errors are gone, but is what I did right ?
Who knows? Why do you think you need so (so!) many wire bends? What do you get if you are running it? What is your loop rate? What is your ptbypt history? So many decisions to make!
Notice that the FFT output is complex, but you are only charting the RE part. If the signal is out of phase, you won't see anything interesting.
At this point, sit down with some nice book about signal processing, fully understand FFTs and decide what you actually want to see.
@Louay02 wrote:
Hi
The errors are gone, but is what I did right ?
That does calculate the vector amplitude.
Is this right from a vibration analysis point of view? I really don't know.
If the movement happens on one axis (not x, y, z but a random vector), using the amplitude would be correct. An improvement to using just x, y or z, AFAIC.
It might still be wrong to use the vector amplitude. Note that the vector could be rotating, and the amplitude will then stay constant. So the FFT won't show any frequency (well, DC). Unlikely, but exemplary for what could happen in practice.
You'd have to find (or become) an analysis expert to get this answer.
Intuitively, I'd say 1) analyze in all directions, or 2) analyze in the most significant direction (not x, y or z, but a projection onto the most significant plane) or 3) project onto a given plane that is of interest.
3) might result in using x, y or z, iff the sensors are carefully placed. That makes this entire discussion mood 🙄.
If you project on a plane, you will get a vector amplitude and phase... That could be used in a complex fft.
This is stuff I'd normally ask my customer, who's usually (not always) the domain expert.
Search LabVIEW like a graph! 