Vibration Measurement using accelerometer

[panneer.s@ssautomation.in]

Hi all,

         We are trying to measure vibration of spherical roller bearing at radial direction with constant speed 700 rpm. Our system has NI-9232 module, 352C33 PCB Accelerometer with 100mV/g Sensitivity, LabVIEW 2021 with Sound and Vibration Toolkit. We want acceleration data to be converted to velocity(microns/s) and check whether the velocity is within limits in three bands(Low band(20-120Hz),Medium band(120-700Hz),High band(700-4000Hz).For that we are applying band pass filters to acceleration then integrating it to velocity then applying FFT to the velocity.

       We found out that High Band amplitudes are much less than expected even though we are using magnetic mounting.(This was verified using Standard Emerson Equipment). Its not possible to do stud mounting in our case.

I have attached VI, image of VI and settings of my DAQ. 

I will be grateful for any suggestions

Thanks in advance.

Panneer Selvam

[Henrik_Volkers]

A short citation from ISO 5348 "Mechanical vibration and shock — Mechanical mounting of
accelerometers":

The quality of the surfaces of the structure and the magnet influences the frequency response of the transfer. A thin oil film can improve the motion transfer.

(In other words, keep it flat, keep it clean and add a drop of oil. Think of adding a spring dampinig system between sensor and stucture if a scratch or dust part is inbetween the surfaces.)

I would "simply" do one FFT on the acceleration signal (in SI units 😉 ) and do integration and frequency band selection in the frequency domain.

[NIquist]

Make sure your accelerometer sensor is as small and light as you can get.  If you can't bolt it down or screw it in, can you glue it on?

[panneer.s@ssautomation.in]

Thanks a lot for your support. I have already tried integration and frequency band selection in the frequency domain with magnetic mounting but still didn't get high frequency signals in medium and high band. I will try the methods you have mentioned for mounting and run the VI again. Can you elaborate on the spring damping system you mentioned? Is it between the sensor and bearing surface?

I am also doubting the sensor. Is 100mV/g sensitivity is enough for capturing high frequency signals?

Thanks in advance.

Panneer Selvam

[panneer.s@ssautomation.in]

Thanks a lot for your support. I think the Accelerometer(3532C33) we are using is small and light.

But in our setup we are trying to check the quality of manufactured bearing one by one, so adhesive mounting is not possible. At first we were measuring vibration with point contact using stud, then we tried magnetic mount(Both flat and dual rail) still we are not clear how to pick those high frequency signals.

I will be grateful for any suggestions. Is there anything needs to be done on software side to improve picking the high frequency signals?

Thanks in advance.

Panneer Selvam

[Henrik_Volkers]

Do you have a flat surface on the bearing? How is your magnet shaped? Size of bearing?

On a blank bearing one would need a magnet for a round surface like PCB Model: 080A133 ..  size matters 😉

My suggestion is to ask PCB /model shop engineers for  mounting solutions, I'm more in the accelerometer calibration part 😉

Maybe not the case, but long cables with a low IEPE current (2mA) limit the slew rate of the signal. (resulting in a low pass filter like nonlinearity) , quick checked by comparing with a really short cable.

For the high freq. damping: Every material between your Sensing element (piezo in the accelerometer) and your source (bearing surface) act like a spring , more or less stiff.

A small crumb or scratch on the surface between this two elements act like a weak spring.

The bearing surfaces i know have a good finish, the dual rail magnet (of propper size and low mass and really parallel rails) with a thin oil or grease layer should do the job.

EDIT: Another method you should try is beeswax (or alternatives made for this propose, again a thin layer)  followed by a mounting pad shaped to the surface and with thin double-sided tape. Due to lower mass than the magnet pad you improve the high frequency response.

copied from https://blog.endaq.com/accelerometer-mounting-best-practices-for-vibration-measurement

(can't use the ISO diagram)  and depending on the publication (well, and application) beeswax is number 3 to 5 (first try wax without mounting pad, sensor only)

[Henrik_Volkers]

Here is another source

https://www.maintworld.com/Asset-Management/Accelerometer-Selection-and-Mounting-Considerations-for-...

If you do some paper research, you find different results. It all belongs to the individual situation (sensor mass , pad mass, surface)

So my suggestion: Use a short cable, clean the surfaces, a little beeswax ball below the bare sensor, support the cable with tape on the bearing (you don't want cable strain), firmly press the sensor on the bearing (flat to bearing should give a line contact),  , give it a try 🙂

[panneer.s@ssautomation.in]

Thanks for your valuable suggestions. I checked and found that we do have a long cable. Planning to shorten it in the next test.

We are planning this test with adhesive mounting(two-part epoxy resin) and stud mount. Since, we cannot directly on the bearing surface(130mm diameter). If this test is also negative as far as HF is concerned, we have to think about the sensitivity parameter.

Will update the results once we are through.

Hoping for the best.

Once again many thanks to your suggestions.

[NIquist]

It would be more expensive, but have you considered a laser vibrometer or con-focal microscope to read the lateral motion of your part optically?  This has the advantage of non-contact, so you don't need to mount the sensor.  No loading of the part and quicker setup will allow you to test large quantities more quickly increasing your throughput.  

If you haven't already, you might also check into NI's toolkits for this kind of thing.

LabVIEW Sound and Vibration Toolkit Download - NI

It doesn't show the license price on the download page but I'm pretty sure this one is NOT free.

[panneer.s@ssautomation.in]

Thanks a lot for your valuable suggestions. We have considered a laser vibrometer at first but due to our budget we can't go for it.

We have already purchased the LabVIEW Sound and Vibration Toolkit for measuring the vibration with the accelerometer(352C33) with 100mV/g. We don't understand why the high band amplitudes are not picking up even though we have tried different mounting types. 

I will be grateful for any suggestions regarding the usage of Toolkit blocks to improve anything in the program side.

Once again many thanks to your suggestions.