How to use STEVAL-MKI208V1K to collect vibration data of bearings?

[HUANG2]

If I use STEVAL-MKI208V1K to collect vibration data of bearings, do I just need to fix this board on the bearing where I want to take data?
Thanks

[Smithson]

I take lots of accelerometer measurements in lots of different places,  if there is a chance of differential movement from the bearing I would stick it down.  A picture or sketch would help.  

Some of our people fix and some don't. 

[Andrew Neil]

---

---

STEVAL-MKI208V1K is just a bare board - so not suitable for permanent mounting, and/or use in an unprotected environment!

Obviously, it will need to be mechanically attached so that the vibration of the bearing will be "felt" by  the sensor!

[Smithson]

STEVAL-MKI208V1K is just a bare board - so not suitable for permanent mounting, and/or use in an unprotected environment!

You can create a small box for it.  I have several in museum. 

It will stick inside an electrical box on a bridge in Iowa with the right tape, and then record through really cold weather.  

If it is a large bearing and the device is sitting flat, it will work in a quiet environment.  

They are actually more robust than they look, until the rubber band that you use to stick somethings together breaks.  

That happened with Flint mounted on the gunboat Philadelphia in the Smithsonian Museum, reliable as heck for years and then dead.  PI finally said, rubber band broken.  I suggested a simple fix, a new rubber band. 

All our accelerometers have names, much better than numbers.

A complex system designed from scratch never works and cannot be patched up to make it work.

True, most of the time. 

[HUANG2]

Hello,
If I add a copper column to the evaluation board as shown in Figure 1 and connect the copper column to a magnet, and the magnet is attached to the red box on the test bench in Figure 2, can this operation collect vibration data normally?
Thank you for your reply.

[Smithson]

Inside the ZIP file is a picture of the sort of analysis you will need.  

If I was doing it the work:

1. Run the accelerometer in an absolutely quiet environment for a few hours, and then do a statistical analysis of the difference between the Gaussian and non-Gaussian component of the FFT magnitude.  You do not need PSD and all the other stuff, you are looking at the pure math of the FFT.

2. I would do the FFT in blocks of 16384 units, so about 8 seconds each.  At 2000 Hz you get FFT to 1000, you can go higher but I doubt that set up needs it. 

3. This will tell you what vibrations you have from the copper extensions,  I would also sit the copper onto a solid surface like a concrete floor and you will get a better idea.  I am doing a concrete block at the moment and this is the difference between the block and the ground.  The best results are counts not amplitudes, but you may not have the time.  The picture is ten days of quiet listening. 

4. Then turn on the machine, you should be able to notice the difference.  It will be due to the complete mechanism so a simple FEM will help.  Really the unit will start to break down immediately it is the df/dt you want. 

5. I would also sit the plain accelerometer on top of the unit and tape it down and just get that record in silent mode. 

6. Thermal vibration tells you a lot.  

It helps if you have the right software, but ours is bespoke.