Repeatable Reading variation in Noise & Vibration Analysies

[BLAK]

We are testing the Quality of the gears based on the noise and vibration produced when the bevel and the pinion gears mesh each other. We Using SV DAQ NI PCI-447 and connected accelerometer and microphone for 1st and 4th channel respectively. We are following signal harmonics method for the testing purpose. First, we acquire the data signals of both noise and vibration and process them by the same above method and judge wheatear the gear has some defects or not. (Based on the threshold values) The problem is, each time when we run the machine, the noise and vibration signals which we acquire will be different due to this final results (Harmonic values) are also changing. We need some filters or any other solution so that we acquire noise and vibration only for gear mesh. I uploaded a picture showing how we doing the test process for sound.

[AlanArmstead]

If you are only concerned about peaks within a certain frequency range, you should adjust the Search Range in the settings of the Peak Search Express VI (on the Configuration tab).  That will tell the express VI to only return the peaks in the frequency range that you choose.

To clean up your results, you may want to try some FFT averaging.  This might help if your peaks are jumping around a little bit due to a low peak threshold that might be picking up some noise.

If you really would like to do some filtering, you could use a highpass filter to get rid of any low order (1x/2x/3x, etc) vibration that you might not care about.  From your picture, it looks like you have the Sound & Vibration Measurment Suite and you are comfortable using express VIs.  So, you could use the filter express VI from the palette here:  Sound and Vibration>>S&V Express Measurements>>Processing>>Filter.

[Ian_Ren]

If you have a speed/tacho chan, you can cut out the gear mesh order directly.

[BLAK]

Dear Alan,

Thanks for your response, we are using averaging method but when we run the machine for the second time the resulting values of averaging are totally different from first time average values.

[BLAK]

Dear Ian,

Thanks for your response, We are using proxymity switches to capture speed & S V DAQ 4471 for Noise & Vibration. can you please explain more on speed / tacho channel thanks

[AlanArmstead]

Could you clarify what you mean when you say the results are totally different?  Could you post some screenshots maybe?

Also, make sure that when you begin a new test, you are setting the "Restart Averaging" boolean input to the Power Spectrum express VI to "True".  Otherwise, it could be buffering some old data and averaging it with your newly acquired data.

-alan