Hi,
I just started working on vibration analysis, that's why I've got a pretty general question.

I am currently trying to analyse machine spindle frequency spectra in order to find out bearing faults...
I calculated all the bearing frequencies (BPFO, BPFI, BSF, FTF) and usually find peaks of one bearing (1xBPFI and harmonics, 1xBPFO and harmonics) at every machine (I measured 5 machines of the same type). They all are pretty new (1 year old).
Does finding these frequencies mean, that all machines are broken or is this just a characteristic spectrum, which does not indicate anything at all?
(The acceleration I measured is about +/-1.5G)

Thanks

Tim

Making an assumption: spindle speed? but not slow & 2kHz and your +/- 1.5 g's P-P? TWF would be OK on general terms.

Please provide more info.

There are many variables such as bearing type, shaft speed, transducer mounting method and distance from bearings, parameter settings, etc. which can influence your overall g's. Having a family of machine tools (to allow for comparative analysis) is a nice advantage to utilize. As far as bearing defect frequency peaks, you will always find some at some level if your collection parameters are optimized and your practices are sound (usually ball spin if nothing else). As you gain more experience, you’ll find that trending, over time, will be your best indicator of the degree of severity. Also, root cause bearing failure analysis can be a very valuable tool to enhance your understanding of vibration data.

Thanks Sam,
I totally forgot the spindle speed! We're running the spindle at 6000rpm. The peak to peak is 3g...
and where sampling the data 100kHz.

It should also be noted that your overall g's could potentially be dominated by imbalance... can you post a spectrum?

Again, not knowing all - I would place the levels as safe. I would basically think the classification would be Cat 1 brg flaws.

Since this is new and also you are a little new: I would step-up my monitoring frequency and see how stable the trend is. If it is consistantly ~3 g's them increase the interval to where you are comfortable. When the levels start increasing, increase your monitoring frequency. It'll come together over time.

thanks for all you comments...
I didn't expect to get comments so quickly

Here's a typical signature (taken at 6000rpm)...
There's minor imbalance, and a lot of bpfo (1x = 1192Hz), some bpfi (1x=1401Hz) and some bsf(1x=575)

Tim

Assuming that your transducer is stud mounted and close to the bearing (minimizing signal attenuation), this signature is typical of a bearing set that is “less than perfect”. In other words, it has minor surface defects present on the bearing components: denting, skidding, etc. that would probably only be visible with a microscope. Remember, trend is your friend.

Thanks,
I was guessing something like that, but I wasn't sure if that was the case.
Maybe you could help me with another question, which came up in my mind today:
Assume, that I've got a machine with a huge imbalance. Is it possible that the imbalance on the spindle excites the ball passing frequencies, due to a different load on the bearing balls?

Have a nice weekend :-)

Tim