Lathe spindle repair, spun bearing, lost preload

Lathe in question is a Victor Vturn 36, which is really a carbon copy clone of an old Mori SL-35. It's a 1997 machine, good working order, but not worth a whole lot.

Noticed that we were getting bad finishes on faces, and discovered some axial play in the spindle, which amounted to a couple of thou, and was somewhat cyclic.

Got a tech in, who dicked around with it for a day and decided he wanted to pull the headstock and send it off for refurb. Quote to do so was right on the edge of economic sense, given the value of the machine.

So I pulled the spindle, and discovered that the angular contact pair had lost their preload, allowing the rear most one to spin in the housing. This in turn had worn the face of shoulder in the headstock casting (circled in red) to some degree, although it seems that most of the wear is actually on the shell of the bearing. There is no apparent wear to the diameter of either the housing or the bearing shell, so I guess it just wasn't a very tight fit in the first place.

I'm thinking poor manufacturing QA is probably the reason for this in the first place, as the shells should not have been able to spin if they were properly retained, regardless of preload.

Anyway, I'm considering options for repair. I have a new angular contact pair on order, I'm leaving the other bearings alone. The front roller bearing is taper mounted on the spindle with no easy way to remove without induction heating and/or sacrificing the bearing. There is nothing wrong with it, so I'm leaving it alone.

Problem is, the cap/labyrinth assembly is captive on the spindle until the front bearing is removed, which is a pretty horrible design IMO. Ideally, I'd be removing some material off the face of that part (circled in blue) to retain the bearing shells properly. I'm considering ways of getting that part in the lathe without removing it from the spindle right now, which is a bit tricky.

Other options that occurred to me are to loctite the rear of the bore, and fit the spindle with shims under the cap, wait for the loctite to cure, remove the shims and tighten it all up.

Thoughts?

The loctite idea should be OK. Maybe 680 which is pretty viscous would be good to "fill" the wear area on the shoulder.

If I was facing the same issue, I'd figure out the new stack height of the bearing vs the bore depth and plan on adding a shim to retain the outer races. Make a shim that is .08-.1mm more than the difference in bearings and bore. Whether it is placed between the angular pair and outer race of the roller or between the outer race of the roller and the labyrinth plate doesn't seem to me that it would matter. That would give a positive squeeze up of the outer races. Be sure to verify that the o-ring that seals the labyrinth ring face will still work and that it has not been squeezed down and set. I'd just plan on fitting a new one.

By any chance did you find the screws holding the labyrinth ring to not be very tight on disassembly? I have seen similarly designed machines where possibly from a crash those screws were pulled or stretched or somehow loosened which relieved the preload on the outer races.

Vancbiker said:

The loctite idea should be OK. Maybe 680 which is pretty viscous would be good to "fill" the wear area on the shoulder.

If I was facing the same issue, I'd figure out the new stack height of the bearing vs the bore depth and plan on adding a shim to retain the outer races. Make a shim that is .08-.1mm more than the difference in bearings and bore. Whether it is placed between the angular pair and outer race of the roller or between the outer race of the roller and the labyrinth plate doesn't seem to me that it would matter. That would give a positive squeeze up of the outer races. Be sure to verify that the o-ring that seals the labyrinth ring face will still work and that it has not been squeezed down and set. I'd just plan on fitting a new one.

By any chance did you find the screws holding the labyrinth ring to not be very tight on disassembly? I have seen similarly designed machines where possibly from a crash those screws were pulled or stretched or somehow loosened which relieved the preload on the outer races.

Click to expand...

Thanks Vanc, was hoping you'd respond. You're pretty much thinking along the same lines I have been, which is reassuring!

The bolts on the labyrinth cap were tight, and there was no crash, and it seems it was hard up against the face of the headstock, so I guess there was little or no interference on the stack length in the first place.

The failure became apparent during a short run of aluminium parts, so the spindle was running relatively fast at the time. 15" chuck machine, 4" spindle bore, 2500rpm max, I guess we were doing a fair bit of running at around 1500rpm, but not more.

As it stands, the stack length measures 0.05mm less than the bore depth, but I will remeasure that with the new bearings before I do anything.

Is your opinion that the labyrinth cap should have an airgap between it's face and the face of the headstock, in order to fully clamp down on the outer races? I'm thinking it must be before you're talking about .1mm shims... It makes sense to me that it should be that way, but it doesn't seem that it ever was originally.

Yeah, do it with shims, you don't want Loctite holding things in there when you have to take it apart, just once more

gregormarwick said:

.......Is your opinion that the labyrinth cap should have an airgap between it's face and the face of the headstock, in order to fully clamp down on the outer races? I'm thinking it must be before you're talking about .1mm shims... It makes sense to me that it should be that way, but it doesn't seem that it ever was originally.

Click to expand...

Having a gap between the faces prior to tightening the bolts is how I was taught, as well as just making good sense. The outer races of the bearings must be captured. Once tightened, the gap will typically close up to where one might get .02mm feller gage to slip, but most cases it will not go in. Lots of factors come into play whether or not that happens. As spindles and bearings get smaller then the initial airgap would typically be less too.

gregormarwick said:

So I pulled the spindle, and discovered that the angular contact pair had lost their preload, allowing the rear most one to spin in the housing. This in turn had worn the face of shoulder in the headstock casting (circled in red) to some degree, although it seems that most of the wear is actually on the shell of the bearing. There is no apparent wear to the diameter of either the housing or the bearing shell, so I guess it just wasn't a very tight fit in the first place.

I'm thinking poor manufacturing QA is probably the reason for this in the first place, as the shells should not have been able to spin if they were properly retained, regardless of preload.

Click to expand...

Your spindle design is described in detail by NSK in the following PDF:


www.nsk-literature.com/en/machine-tool-spindle-bearings/offline/download.pdf

Take a close look at page 44. The page is titled " Mounting the heavy duty spindle"

NSK states that the angular contact bearing pair behind the double roller bearing is designed only for axial loads. This is a not the standard angular contact bearing design. There is not a OD interference fit for this bearing pair. The most likely cause for the loss in preload is the lock nut backing off. It looks like there is a set screw to prevent this from happening.

Robert R said:

....There is not a OD interference fit for this bearing pair. .....

Click to expand...

Typically there will be a few micron radial clearance for the outer races.

The outer races of a thrust angular bearing set must be captured axially with some amount of preload. Without that, the spindle assembly will have axial float.

Robert R said:

Your spindle design is described in detail by NSK in the following PDF:


www.nsk-literature.com/en/machine-tool-spindle-bearings/offline/download.pdf

Take a close look at page 44. The page is titled " Mounting the heavy duty spindle"

NSK states that the angular contact bearing pair behind the double roller bearing is designed only for axial loads. This is a not the standard angular contact bearing design. There is not a OD interference fit for this bearing pair. The most likely cause for the loss in preload is the lock nut backing off. It looks like there is a set screw to prevent this from happening.

Click to expand...

Hi Robert,

I hadn't seen that document, but have been busy reading this one, which contains more or less the same data, and is perhaps more pertinent to the specific bearings I have here (BAR series). The exact bearing is 160BAR10SDBELP4.

You are correct that the outer race is intended to have radial OD clearance, however I can confirm that the preload was lost due to simple wear and tear, the nut had definitely not backed off, as it's rotation was marked up with paint marker from installation. The lathe is 22 years old after all. We have had it for 10 years and it has worked pretty hard in that time, but has not been mistreated, and never crashed, at least while we've had it.

The very first thing we did was try and tighten that nut, and while it did tighten very slightly more, it didn't affect anything. After I pulled the spindle I messed around with it some more, and was unable to tighten it such that the outer races of the angular contact held against each other.

I'm pretty convinced that the axial stacked interference was incorrectly set from the factory, providing insufficient clamping force retaining the outer races. Then, a combination of the the preload being lost after 20 years of use, and a period of relatively high speed running warming everything up and perhaps expanding the housing, allowed the outer races to move. I could well be wrong, but that makes sense in my head!

So I got around to this finally, and the lathe has been running without problems for the last week or so.

Following Vanc's and Hu's advice I fitted the new angular contact pair to the spindle and measured the stack length; It was 0.15mm shy of interference, so I made a 0.2mm shim and fitted it between the races of the front roller and the first angular contact bearing.

I had a bit of trial and error getting the rear roller bearing preloaded appropriately as it is fully adjustable, I overdid it on the first go which caused it to warm up quite quickly during run in, but backed it off a touch and all is well.

Curiously, the lathe always had a bit of taper which I attributed to the headstock needing aligned. It wasn't a big deal so for the parts we were making on there, so I never did anything about it. The guy who's been running the machine for the last couple of days insists that it's turning bang true now. Given that it was only the thrust bearings that failed, I can't get my head around why that might be...

I have successfully rebuilt four different CNC lathe spindles here at the shop over the years. (Belt driven, I would never tackle an integral spindle rebuild in the shop).

Although, I never really know if a rebuild will work until it’s all back together and run for a while, as I’m not quite as particular about all the measurements and stack ups when rebuilding.

I basically measure the existing bearings, and machine spacers so the new bearing-stacks match. Then torque the hell out of the lock nuts, spinning the bearings by hand to insure I don’t get them too tight.

Of course everything has to be kept perfectly clean, not over-greased, and handled gently when re-installing.

It feels good to be 4 for 4!

ToolCat