Low speed bearing monitoring


I enclose an article from SKF regarding the low speed vibration monitoring where they report that it is better to use an accelerometer of 100 mV/g and not 500 mV/g; however there are several articles on the Internet where they indicate that it is better to use an accelerometer of 500 mV/g and not 100 mV/g; I would very much appreciate your opinions on the matter regarding the SKF articles in view of the fact that this issue has generated a lot of controversy in the personnel that monitor the low speed bearings.

Greetings and many successes.


Original Post






How about a different technology or different/additional way of verifying a fault exists. The above links talk about ultrasound for slow speed bearings, can be route based or permanently mounted.

As for 100 vs 500 mv/g.....as long as the low end/high end of the sensor is capable of reading frequencies below a certain speed or up to a certain frequency, either will work. Stud mount. You will most likely not see 1x 2x 3x rpm depending on shaft speed


Bearing analysis with SKF equipment utilizes acceleration enveloping(gE) technology.  When a bearing is damaged, it excites the bearing natural frequencies.  Acceleration enveloping has 4 band filters to choose to capture this bearing natural frequency.  Filter 1: 300cpm to 6,000 cpm; Filter 2: 3,000cpm to 60,000cpm; Filter 3: 30,000cpm to 600,000cpm; and Filter 4: 300,000cpm to 2,400,000cpm. 

Filter 3 is usually the most sensitive. Excited bearing natural frequencies are usually in this range.

The 500mv'g sensor frequency response rolls off on the high end well before the bearing natural frequencies and therefore useless in bearing detection even at slow machine speeds.  The frequency response of a 100mv/g is well within the range of the bearing natural frequencies.



Bear in mind that as the bearing defect progresses, it will show more peaks in its other spectra (500 mV/g) and less in its HFD spectra (100 mV/g). If you find a defect in the bearing in the HFD spectra (100 mV/g) and not in its other spectra (500 mV/g), this is a problem that you must monitor. If you have a floor with high noise in the HFD spectra (100 mV/g) and defects in your other spectra (500 mV/g), you should replace the bearing.
Therefore, the two accelerometers (100 mV/g and 500 mV/g) should be used so that no detail escapes; also with the accelerometer of 100 mV/g will not be able to detect anomalies at 1X, 2X, ..depending on shaft speed.

Due to the band filter range of the filters of the software, data will not be collected in the low frequency vibration range (1,2,3 x RPM) as well as the very high frequencies.  If you are collecting data in the filter 1 or filter 2 range, you can get away with using a 500 mv sensor due to the upper frequency levels of the sensor.  If you are using filter 3 or especially filter 4, you MUST use a 100 mv transducer to collect the data due to the upper range of the sensor.  A 500 mv upper frequency range will not go out to 10 KHz needed to collect a filter 3 reading.  Filter 4 is measuring out to 2.4 M cpm and is obviously way outside that operating range of a 500 mv sensor.

We have had great success identifying bearing defect frequencies down to 3 RPMs.  It is very possible with the correct settings.  Identifying the frequency range where the bearings are indicating a problem is the key.  For example, have identified 11 RPM kiln trunion bearings using filter 3 and a 100 mv sensor a year in advance of stage 4 failure.

Filter ranges are as follows:

Filter 1 = 300 -   6K cpm          Filter 2 = 3K - 60K cpm

Filter 3 = 30K - 600K cpm        Filter 4 = 300K - 2.4 M cpm

Add Reply

Likes (0)