vibration/alignment/balance

I have a motorized spindle (synchronous motor) on a machining center running at 15,000 rpm. A time waveform showed "beating" and the spectrum showed a peak at 15,000 rpm (0.9mm/s rms) and also a larger peak at 16,500 rpm (1.3 mm/s rms). There were also peaks at 2x, 3x and 4x but all small and less than 0.1mm/s rms. The rest of the spectrum was clean. This is 1 of 4 identical machines but the only one with beating and these 2 distinct peaks. What could be producing the 16,500 rpm peak?

In some designs bearings have pretension one way or another to reduce bearing play to minimu and a standard problem that will give accuracy problems sooner or later is that the pretension thingy is exhausted, broken or something loosened up and it normal give looseness pattern of 1xRPM multiples, my guess is that this then excite some resonance at the odd freq. but may be wrong. You may also have something inside turning where it shouldn´t. A thing that looks like a tube may include some magic mechanics inside, once you see it taken apart or the drawing.... and some is real simple. Some designs I still can´t realize they can work and still they do for 20 years of daily use. On the machines I have seen this deterioration of pretension is a much more common cause of bad operation than common bearing faults. Bearing faults come with failed seal or a crash, here they say if you accidently knock off a thing of the size of a M8 bolt thread you have most certainly broken the bearing, lees than that it makes it most of the time.... So if it is of a kind where pretension can be checked try to do that. Otherwise send it for service that´s only btw. 3 and 30 KUSD or so. So it´s not unusual they run until it produces bad parts "to make the most of it". I think one reason for the wear problem is that they stuff the machines with tools that are many times larger than the spindle was made for so they get short life length of the spindle parts, but that´s what I think not the tool makers. Olov

Thanks for the feedback. In response to some of your questions, the parts produced do not have chatter and meet all form and finish requirements. The spindle appears to be in good condition as evidenced by the parts produced and by the spectrum. There is only a tiny peak at 0.46x which is the FT and nothing at 15.2x (the BPI) or 12.8x (the BPO) or at 6.5x (the BS) or any multiples of those frequencies. The spindle does have ABEC 9 bearings and is at the correct temperature. The BCU reading is 1.3 and there are no harmonics in the demodulated spectrum. Apart from this peak at 16,500 rpm there is nothing "unusual" about the vibration signature. There is a balanced tool holder in the spindle and the draw bar is engaged so there shouldn't be anything loose inside but why we get this peak is still a mystery.

I usually take data with empty spindle to make sure to get repeatable measurement, do you have data from both front and back brg? Is it direct drive, no gbx or otherwise? Some times full speed is not the best to take data due to resonances although it´s GM procedure but it shouldn´t produce freq´s by it own. Have you checked at another speed or a run up/down so it follows the machine RPM or not? You don´t have a peak at 13500 RPM also if you look for it so there is some kind of modulation? So last resort, does it disappear direct when you cut power (if you can) so you can blame the electrics? There is a lot of those electrons around such machines nowadays. Olov

taking data with "empty spindle"
Some drawbar or tool retention systems emphatically prohibit running fast without a tool in place.

Dan T