I am using a triaxial ICP acceleromoter (PCB model 356A01) connected to NI 9234 on a cDAQ-9174 chassis. The accelerometer will be used to monitor vibration on a running AC motor and will be compared against a third party vibration analysis machine (CSI 2120). The goal is to compare measurements in units of inches per second on both devices.
I'm using the DAQ assistant and I believe I have everything correctly set (continuous, 25.6k samples, 51.2k rate). My issue is after integrating acceleration (to get velocity) the data looks wild and I don't know what's going on. It also appears inconsistent see Pic1 versus Pic2. These pictures are on different days but with the same DAQ settings under the same testing conditions.
Another thing, I'm not certain if SVL integration VI expects units in g's or m/s^2 on the input side as its not specified in the help document. Pic1 is the new norm. For comparison, the CSI 2120 shows a reading of 0.0514 in/sec that's fairly consistently on the exact same spot. Please check my VI to see if there's any errors and I'd appreciate any tips on interpreting the data I'm seeing.
Solved! Go to Solution.
Don’t have LabVIEW at the moment, but looking at your pictures, it looks like you have some low frequency drift/noise. Try using a low pass filter before you do the integration.
Actually, what mcduff *meant* to recommend was a *high* pass filter.
First: A happy new year 😄
I strongly recommend to first convert and calculate in SI Units 😉
in the SVT scale the input in mV/(m/s²) ( if you have a ISO 17025 calibration certificate, if must be noted there;) )
use the SVT integration vi and activate the high pass filter (in the range or lower of the ICP high pass filter)
Result is in m/s and can be scaled to the imperial units if needed.
Happy late New Years to all!
Sorry for getting back so late I haven't had much of a chance until recently. I did as you said Henrik and feel it is closer but not quite there just yet. I'm not certain what the "ICP high pass filter" refers to but I do see a frequency range of the accelerometer (says 2Hz to 8000Hz on x & z axis, 2Hz to 5000Hz on y axis). So I put 2 into the highpass frequency cutoff for the SVT integration. Still can not get it to show similar results as the stand-alone vibration analysis meter. I appreciate the responses thus far thanks guys.
I am not sure what you are doing or exactly what you want, but your two pictures have way different frequency responses. One picture seems to have a strong 50 Hz component as indicated by a ~20 ms period, whereas in the other picture it is missing.
Where are you located? Is it possible you have 50Hz line frequency pickup in your system?
Looking at the VI you posted with the data, I see no 50 Hz signal in the data, see below.
I'm trying to recreate the functionality of a vibration analysis machine. Right now I'm trying to nail down the overall velocity in units of inches per second. I'm comparing the results I see from the National Instruments equipment to the handheld device. See pic4. I know if I see similar readings I'm probably close to being where I should be. As of right now I'm not sure what should be the high pass cutoff frequency (for the SVT integration VI). I believe it was supposed to be 2Hz but I tried your 100Hz that I saw above and it works much better. Any reason why you chose that number?
I'm not sure about the 50Hz pick up but I did notice a tremendous amount of noise in the cable. I'm in south Texas. The accelerometer's BNC cables were very long and on top of the test motor's leads. I reduced the length and put some distance between them and got the following results. See attached VI.
Again not sure about the highpass frequency but it seems to be working for now. Thanks.
Also if I’m not mistaken… I’m assuming the peaks in the velocity graph are the overall velocity that I’m looking for. I think I would detect the peaks and then maybe averaging them though I’m not too certain about that
I believe it was supposed to be 2Hz but I tried your 100Hz that I saw above and it works much better. Any reason why you chose that number?
Did you see my previous post? I took the FFT and looked at the frequencies present in your data. There was no low frequency data in your sample, you could go higher than 100 Hz.
A few points: