High speed vibration analysis issue

[webcmc]

You're summary of what I'm attempting to do is correct. Basically, we want to acquire data for a set amount of time (say 30 secs) at different rmp set points.

For example, we would ramp up to say 250rpm, take 30 secs of data (3 channels) and determine if the vibration is out of acceptable limits. Then we would ramp up to 1000rpm, take 30 secs of data and do the limit test. We would repeat this for a series of rpm set points. We also need to log the resulting fft data so engineers can look back and see what was going on if something when wrong.

We were planning to combined the three axis of data using:

sqrt(sq(ch 1 array) + sq(ch 2 array) + sq(ch 3 array))

then run the resulting combined array through an FFT. Then do the limit checking and logging of the resulting waveform.

Not sure if this is possible or even the right thing to do. (Which is why I started this thread 🙂

Hope that clarifies what we're looking to do.

- Chris

[webcmc]

I think I need to see an example of this as I'm getting a bit lost. Is there one I can look at?

[webcmc]

Attached is a version based on the non-continuous daq using a simulated PCI-6250 card.

One thing I don't understand is if it's set to collect 720,000 samples @ 24kHz it should take 30 seconds to collect (or simulate 30 secs to collect), but it does this in less than 1 sec.

Also, is this what the fft graph is expected to look like based on this data? Something doesn't seem correct.

- Chris

[Thaison_V]

Chris,

With a simulated DAQ board, the points are generated right away, so it will not actually take the full 30 seconds to acquire the points.  With the actual hardware, it will take at least 30 seconds to acquire.  The simulated data is a sinewave with a little bit of noise, and from the FFT results that I am seeing, this is what I would expect.  There is a spike at the frequency of the sinewave and the rest is noise.  Are you expecting something different?

Thaison V
Applications Engineer
National Instruments

[webcmc]

Nope - that's what I was expecting. It was the timing that was throwing me off.

How can I adjust the frequency of the sinewave produced by the simulated board?

(I'm pretty new to the whole test and measurement world so many of these questions are probably trivial 🙂

- Chris

[preston_johnson]

Here is a simple example,

I used a usb-9233 to acquire two channels of vibration and one tachometer channel.  I configured the DAQmx express VI to acquire data.

I wired the result of the acquire data VI to the FFT express VI and set averaging to 30 averages.  I am collecting data at one block per second. 

The loop terminates when the 30 averages are done.

I export the data to three waveform data types (really should be a cluster) so that the three FFTs can be limit tested later. 

I have attached the VI

[webcmc]

Thanks for putting together an example. Unfortunately, I can't open it as I'm using LV 8.0.1 and it looks like this example was built with 8.2. Is there any way you could save it in an earlier version that is compatible with 8.0.1?

- Chris

[preston_johnson]

Attached is a bit map of the diagram

1) use daq assistant to set up continious daq for your board

2) use the express spectral vi to take in the signals and configure for 30 averages (if you are getting 1 second blocks of data)

3) terminate the loop when the RMS averaging is done

4) outside the loop split the signals and put the resulting individual FFTs in the structure you need for later evaluation

let us know what questions you have