Hi all,

Can anyone give me an explanation of how to use CPM for roller bearing vibration analysis? or provide a link to where I can read up on it?

We have 2 vibration systems, one is old that I am not very familiar with, it is set up to gather data in CPM's mostly.

With the other newer CSI based system, most of the data gathered can be switched to frequency, orders or CPM's.

Frequency and orders makes sense to me and I can use them.

Thanks
From a Newb
REJ

CPM=Cycles per minute. CPM is related to RPM. If a machine is operating at 1785 RPM and vibration is seen at 1 X rotation it will show in the spectrum at 1785 CPM. 60 HZ = 3600 CPM, 30 HZ = 1800 CPM etc. Bearing faults can also be defined in CPM's. I'm at home right know and my information is not available but maybe someone can provide the formulas for bearing frequency calculations.

CPM = cycles per minute

Hz = CPS = cycles per second

To convert from CPM to Hz, divide by 60. Thus, a motor turning at 3600 RPM is also turning at 60 Hz.

I prefer to think in terms of Hz, just because the numbers are more manageable. Say our 3600 RPM motor is turning a gear with 100 teeth. Gear mesh frequency is 360,000 CPM. Using Hz, it's 6,000 Hz, a number I can deal with without counting all the zeros on my fingers.

Orders is a bit different. The frequencies are normalized to the machine speed. The rotation rate is always 1 order, twice rotation is 2, if the pump has 5 impeller vanes, the vane rate is at 5 orders. Scaling in orders is very useful for comparing data from measurement to measurement where the machine might be running at a slightly different speed each time it's tested.

To convert orders to frequency, just multiply by the rotation rate. If the motor is turning at 3600 RPM, the first order is 3600 CPM (or 60 Hz).

Bearing tables usually list the defect frequency in terms of orders. If a suspected bearing has a fault frequency at 2.3 orders, just multiply by the rotation rate to see where it should show up on the graph. If you have a suspected bearing tone in the spectra, divide the frequency of this tone by the frequency of the rotation rate peak to determine what order it is. Of course, you can use either CPM or Hz to do this and come up with the same answer.

Jon
Spintelligent Labs

Further to the other various replies, here are the formulae for calculating your bearing frequencies.

Bearing frequencies can be calculated precisely using:

S = shaft speed
PD = pitch diameter
BD = ball or roller diameter
N = number of rollers or balls
ø = contact angle
FTF = Cage Fault Frequency
BPFO = Outer Race Fault Frequency
BPFI = Inner Race Fault Frequency
BSP = Ball Spin Frequency (Rolling element fault frequency)

FTF = ½ S x [1 - (BD/PD) x cos ø]

BPFO = N x FTF = ½ S x N x [1 - (BD/PD) x cos ø]

BPFI = ½ S x N x [1 - (BD/PD) x cos ø]

BSP = ½ S x (PD/BD) x [1 - (BD/PD) x cos ø]


Bearing frequencies can be ESTIMATED (±20%) using the following formulae:

FTF = 0.4 x S

BFPO = 0.4 x N x S

BPFI = 0.6 x N x S

BSP = 0.23 x N x S (N < 10)
or 0.18 x N x S (N ≥ 10)

Hope these help.

BPOR + BPIR = NUMBER OF BALLS.

BPIR = 60% NUMBER OF BALLS
BPOR = 40% NUMBER OF BALLS

Handy when you don't know the "exact" numbers.

I dont like CPM. Are big numbers.
Hz is more rational.
Order have some advantages, but is necessarie entry correct value for the rotation.

Ricardo Góz from Brazil

I'm sure it must be one of those personal preference things. I can't even begin to think in Hz. For me CPM is more logical because it relates directly to RPM. Orders are useful, especially when looking for whole multiples of running speed (blade pass, gearmesh, etc). I use Entek Odyssey which can scroll from CPM to Hz to orders without closing the plot.

There are many ways to do many tasks in vibration analysis. Just use what's good for you and gets you results.