Practical Solutions to Machinery and Maintenance Vibration Problems
Chapter 8, Vibration in Bearings
Section 11, Further Considerations Regarding Calculated Bearing Defect Frequencies
A common error is to suppose that most calculated bearing defect frequencies will be present almost always in the higher frequency portion of the vibration spectrum. This, of course, can occur; but most bearing defect frequencies that can readily be calculated, are revealed at frequencies that are relatively low and can easily be mixed up with those from other vibration sources.
For example, sometimes bearings have defect frequencies and harmonics of those frequencies in the range that normally shows the lower harmonics due to shaft/coupling misalignment. If the instrument resolution is not high, accuracy is sacrificed and sometimes bearing defect frequencies are confused with frequencies such as those originating from electrical hum, misalignment harmonics of 2 x and 3 x rpm, vanepass frequencies, and so on.
The calculated frequencies for a defective cage or retainer are almost always in the range of less than 50 percent of rpm. Typically, the frequencies are from a little over 30 percent to slightly under 50 percent of rpm. A surprising number of bearing cage defect frequencies are in the range of 41 to 48 percent of rpm, causing the symptom to look like that for oil whirl. Yet, rolling-type bearings are not subject to oil whirl!
Another characteristic of a cage defect frequency is that the calculated amount (less than 1 x rpm) often does not appear on the spectrum. Instead, a harmonic of that number does (such as 2 x the bearing cage defect frequency). This can come surprisingly close to the 1 x rpm frequency and be revealed only with proper resolution.
For example in Fig. 7, consider the spectrum's first peak at 28.0 Hz and its third peak at 96.2 Hz. At first glance, the first peak looks as if it's surely a ½ x rpm sub-harmonic. Through a quick calculation, it is shown to be a non-synchronous 0.47 x rpm. The third peak appears to be its third harmonic but calculates to be 3.44 x 28 and, therefore, not its full integer multiple (therefore non-synchronous). Calculation indicates it is also close enough to appear to be 1 ½ x rpm.
Also notice that the frequency range goes up to 800 Hz. If 400 lines
were used, the range for each line would be 2 Hz (120 cpm). Reducing
the top frequency to 200 Hz gives 0.5 Hz (30 cpm) per line. Increasing
the number of lines to 1600 gives a resolution of 0.125 Hz or 7.5 cpm.
Obviously, data collecting for routine predictive maintenance requires
fast procedures, but when a bearing or some other defect is suspected,
a further set of very accurate data is required for proper analysis.
This textbook contains only part of the information in our Practical Solutions seminar.