Spindle Bearing Replacement

I am working on replacing or regreasing the spindle bearings on my PM833-TV. I have removed the spindle from the machine and removed the castle nut that sets the preload.

Can someone give me a hint as to how to remove the spindle from the Quill? I don’t want to force anything but it seems like I need to press on one end of the spindle and the nose of the quill? Is there anything else holding this think from coming out?

you might find some clues how it is built if they have a parts diagram in the manual.

I took the spindle out of the machine and here is what I found:

  1. The bearings are nice NTN brand bearings that have a max speed with grease of around 6000 rpm.
  2. I removed the castle nut that maintains the preload, and it was completely loose, it took absolutely nothing to remove it.
  3. When I did my test run with the oil seal removed, I reached a steady temp of 175 on the outside of the quill at 4500 rpm.

My goal was to remove the OEM grease and replace it with a high speed spindle grease to help lower the temperature so I can run at 6000 rpm. I think if it is currently running at 175 degrees at 4500 rpm, then it will run too hot at 6000 rpm. I rewatched the Dr. D video and unfortunately, both the bottom and top bearing are press fit, so I cannot remove then without a press (I do not have one).

I always expected the bottom bearing to be press fit, but one thing is confusing. If the top bearing is also press fit, then how does tightening the castle nut influence the preload. How can it press on a bearing that is pressed to the spindle?

Here is a drawing of the spindle:

I ordered a backup set of SKF bearings just in case. Todd Candee told me he replaced the bearings with SKF, used Isoflex NBU 15, swapped the oil seal for a slightly large one, carefully set the preload and runs for 6 hours at a time at 6000 rpm with a spindle temp of 180F.

I think I need to find a friend with a press. I can’t think of a way to wash out the OEM grease from the original NTN bearings to replace with Isoflex NBU 15 without removing them from the spindle.

Ok… I spoke with the folks at Precision Matthews and they said I can use a soft blow hammer and remove the bearings. I guess I was afraid to hammer this thing… But it worked! I 3d printed some jaw protectors, positioned the spidle on top of the vice jaws, placed a piece of aluminum on top of the spindle, give it a few hard blows with a 2lb dead blow… and done! I used a similar approach to remove the bottom bearing from the spidle.

glad it worked out :slight_smile: i started looking at the exploded view and then seen you made progress.

Thanks Jay. I have no experience with this stuff, so I tend to be very cautious.

When you tighten the castle nut, you are putting the spindle in tension and the bearings into compression. Also, as I am sure you have noticed, its just the outer raceways that are pressed into the quill. If there is no preload, then the inner raceway and bearing cage could move away from their respective outer raceways.

Thanks David. That is exactly what I thought.

I spoke with David Clements from Arizona CNC this morning. He said that the top bearing should be LOOSELY press fit to the spindle. It is supposed to be a loose press fit. However, on the 11 mills that he has personally worked on, several of them had too tight of a fit between the top bearing and spindle. This made it very difficult to adjust preload. When adjusting the preload, it would resist, then move too far. In these cases, he did very lightly wet sand them with 2000 paper. I may have to consider this on my spindle.

One interesting point Dave mentioned is that he sometimes fine tunes the preload by using a dead blow hammer and hits the top or bottom of the spindle to tighten or loosen preload. He does this with the spindle installed in the mill.

Thanks… Richard

Hey David… Quick question. How did you remove the bearing races from the quill?

I used a long bass screw driver to hit the small lip that is exposed on the back of either bearing. For the bearing closest to the nose, I reached through the slot in the spindle that is used for tightening the set screw on the collet and for the back one, I had to go all the way through the spindle.

Thanks David. I was just trying this approach when Amazon arrived with a tool designed to pull bearings.

Amazon.com: Shankly Two Jaw Harmonic Bearing Puller and Gear Puller Universal Bearing Puller Tool or Pullers for Mechanics Heavy-Duty Pilot Bearing Removal Tool Small Bearing Puller : Automotive

I reversed the legs so the talons could grab the underside of the race, 3d printed a riser of sorts to allow me to pull the ring completely out… and BAM… the races popped right out.

I was surprised to find a bunch of oil leaking out of where the race was. Any ideas as to why?


Nice! I didn’t think the lip was big enough to use a gear puller - glad you had success.

I had some fluid leak out but I just assumed it was coolant/mist that condensate inside the spindle and mixed with some of the bearing grease.

Strange. The oil that leaked out looks like way oil. Can’t imagine how it would get there. I was wondering if it was intentionally used to make the race install/removal easier. I watched a video from Robrenz on a precision spindle rebuild. (https://www.youtube.com/watch?v=grUdsTTRGl4&t=4972s) He is very knowledgeable and previously worked professionally rebuilding spindles. He noted that they always applied a very thin layer of extreme pressure grease on all bearing fits and the locking nut face.
(Amazon.com). I was planning on using this but was caught a bit off guard when all this oil leaked out from behind the race. Really not sure what to think. I checked the Machinery’s Handbook but found nothing. I may call SKF to see if they have any guidelines regarding outer race installation.

BTW… I noticed that the splines on the top of the spindle were dry… Did you apply any grease to inhibit rust?

Thanks again… Richard

Yes, I sprayed a little white lithium grease on the splines, but I did not re-install the quill feed mechanism (spring, etc.) as I don’t ever plan to use it.

Ok, I successfully installed the new SKF tapered roller bearings. I have one question regarding preload.

I slowly tightened the spindle nut until all endplay was removed. I measured this endplay by placing a magnetic base on the quill/spindle body and the dial test indicator on the spindle nose. In a similar fashion, I then measured the runout in the spindle. It was immeasurable with using my 0.0005 dial test indicator!!

I then ran the spindle for 30 minutes at increasing rpms and compared the temperature to the OEM NTN taper roller bearings. The spindle was not loaded at this time. (see below).

I am bit concerned. With the OEM bearings, at 4500 rpm it was running 175F. With the new bearings I am running only 93F at 5000 rpms. Several other hobbyists that have the exact same machine with these SKF bearings report a steady temp of about 180F under load at 6000 rpm. SKF says that an operating temp of up to 225F is fine.

Do my lower temperature readings mean I do not have enough preload? When adjusting preload, should I have continued tightening a bit further after the point that all endplay was removed?

I did expect my temps to possibly be about 30 degrees less since I am temporarily using an oversized oil seal to rule out issues associated with a tight oil seal, but my temps are as 84F degrees lower at 4500 rpm.


I would put a long tool in the spindle (with an indicator on the nose) and pull on it to simulate radial load. Anything more than 0.0005” would warrant additional preload. But really, I would always trade a too hot-to-touch tool for a bit more runout in the spindle.

Thanks for your quick response, David. I like your idea for presenting a radial load. I will give it a try shortly.

Ok. I setup the Dr. D test jig (see below). I tugged on the end of a fairly long 1/2" inch drill bit and the dial test indicator moved less than one graduation or less than .0005. This seems too good to be true. Not sure what to make of it. The spindle temps just look too low. I have to think on this one.

Take II. I wasn’t completely satisfied with the previous test, so I found a 3/4" one foot long steel rod and installed it directly into the R8 Collet (see below). I pushed and pulled radially and found far less than .005 deflection. I also pulled on it axially up and down to check for endplay. Still immeasurable.

I am still unclear on the topic of preload. My understanding is that these tapered roller bearings (like angular contact bearings) are designed to work in pairs. In this scenario, they are designed to have an axial force pressing the bearings together. I believe the ring geometry is such that they act as springs that get compressed by the axial preload. If this is correct, then adjusting the spindle nut just to the point that there is no endplay, would not compress the ring geometry and therefore result in no preload.

Does this make any sense?

Thanks… Richard

The primary purpose of the preload is not to compress these components in the same way a spring would be compressed. Instead, preload is meant to ensure zero clearance. From your thorough investigation, it seems as if you have eliminated play. The only thing I can think that can cause your spindle to run cooler than expected is not packing enough grease, which might initially cause the spindle to run cooler.

Honestly, I enjoy your investigations, so I don’t want to tell you stop, but I would start cutting parts and check tool life and play in the spindle from time to time. The bearings have to be re-greased every 500 hours anyways.

Great work overall though :grinning: