So the upper bearing is a sliding bearing? ... Is it sleeve or it tilt pad?
The lower bearing you described as ball bearing - is that deep groove or some combination of angle contact bearings? And it takes the weight of the rotor?
If the upper bearing is a simple cylindrical sleeve design then Greg's suggestion oil whirl is plausible (although typically it's would occur at a frequency less than 0.5x). Tilt pad shouldn't be nearly as susceptible to that. If it is whirl, it may be sensitive to oil temperature among other things.
Also the bottom bearing can be a source of problem. It may be sensitive to thrust loading especially if there is an angle contact bearing. Then the problem might possibly be lack of external downthrust (yes I realize it didn't vibrate incoming without thrust load, but see my note below (*)). More likely there might be assembly factors which establishes preload of angle contact bearing even before external thrust is applied (how tightly is the angle contact clamped axially).
A little bit on the unlikely side, being a vertical motor with oil lubricated upper bearings you probably have a standpipe forming the inner diameter of the upper bearing oil reservoir, and then some piece (maybe called a runner or collar) which mounts to the shaft at elevation above the standpipe and provides a surface to interface with the bearing stationary components at an elevation below the standpipe. If that piece (runner or collar) is too loose on the shaft (and not well clamped axially) then I believe that can cause non-sync vibration as the collar itself tilts in a whirling pattern with respect to the shaft.
On the far-out unlikely side, if inner ring is spinning on the shaft you might get non-sync frequencies and probably worsening vibration, heating and failure in short order... I assume shaft fits to inner ring were checked during the maintenance?
Also on the far-out side, did you check the FTF (cage fault frequency) associated with the lower bearing part number to see if it is 0.45 orders?
(*) We have one family of vertical motors with rolling element thrust bearings which give non-sync frequencies (not assocociated with any fault pattern) which show only intermittently during uncoupled run, but always go away coupled.
Fwiw, we also have one family of rolling bearing vertical motors which has and unusual non-sync vibration not identifiable as a bearing fault patterns and in this particular case it doesn't go away coupled or uncoupled, but comes and goes for reasons we can't even narrow down. We have never figured out a definitive cause (even with lots of help from our motor shop). After spending a lot of time and money I finally gave up trying to figure it out. It has been that way for decades and never associated with any degradation or reliability problems on those motors so it's not worth any further investigation.
