Bearing Defect Detection

[roryrobb]

Customer refuses validity of acceleration envelope readings to detect rolling element defect frequencies in newly installed bearings.

Specifies no more than 3 peaks in a single spectrum in alignment with a bearing element frequency with an amplitude greater than .0049 i/s.

does this seem reasonably achievable to anyone?

For a 4 pole motor, setting Fmax at 19200 with 800 line resolution.

 

[electricpete]

I am on the customer end of the repair business (we send our motors out for repair).

Assuming the bearings have been replaced (usually the case), I think it is achievable. Our local shop says nothing with a single peak exceeding 0.01 ips at a defect frequency/harmonic goes out the door. The only exception might be 1* BPFO which occasionally shows up and seems fairly benign.

I can understand customer reluctance to embrace demod - the details vary by vendor (and within a vendor vary by filter setup) so how do you know what you're getting.


=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

I think it also depends to small extent on bearing manufacturer. Link Belt is notorious for defect frequencies right out of the box. Apparently some bearings are smoother than others.

The handling of the bearings during shipment, storage and of course during installation can also cause problems. Drop a bearing on the floor and you will probably have defect freuqencies. If the bearing is handled a lot without regard for cleanliness a very small piece of debris/dust can cause a problem.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[roryrobb]

Pete you've been doing this a long time and your opinion is most valuable. Thanks for the response!

I agree except that .0049 is a bit tighter than .01 ips.
It is acheivable, sometimes costing multiple sets of bearings and assemblies. Handling prior to our plant is out of our control, and there seems to be no mechanism for verifying an unmounted bearing other than feel and appearance.

True, in the ideal world, no defect should be present to detect, but at what point does it become uneconomical for the shop or the customer to purchase multiple sets of bearings to acheive rocket science precision?

 

[Tmoose]

EPete said "The only exception might be 1* BPFO which occasionally shows up and seems fairly benign."

Any chance that benign BPFO appears on bearings with bolts clamping a retainer plate axially?

I have caused BPFO (up to 0.3 ips!) by inadvertently tweaking bearing housings in ways that cause the bore to pinch inward or go oval a few thousandths.

When threaded holes are close to a bore ( bearing housings, many engines' cylinders) upon torquing, a bump appears in the bore at each bolt. Iron Harley Sportster cylinders finished "perfectly" round will have a high spot of 0.0002 to 0.0004 inch over each tightened head bolt.

http://i.ebayimg.com/04/!BV1nGcQCGk~$(KGrHgoH-DkEjlLly9hRBKUj13D1yw~~_1.JPG.

Remove the cylinder after a few thousand miles and the piston rings will have worn each bump down to a shiny spot over each each head bolt

The practical solution is to finish the bore only with correct length bolts tightened against a dummy hollowed out cylinder head called a torque plate.