What's new
What's new

Schaublin 135 spindle bearings assembly and preload procedure

I assume the measurement should be taken without the front flange, so access should not be an issue IMO, but I have to take a closer look.
Are these adjustable parallels accurate (stable, finely adjustable) enough for this task when you chasing one micron?

Yes, just the shaft and the roller.

The gauge blocks/parallels; no, not really that accurate, but they dont need to be. You arent chasing a micron here or any particular number. You are measuring that width in a few places, taking an average and comparing it to what you "need" (what the formula spits out). Then if you need to be thinner, get to grinding, if it needs to be wider you can make a new spacer or shim the old one accordingly.
After a while of leaving the spindle aside, equipping myself with some relevant tools and with a help of a friend I attempted to make the first step of measuring the front bearing internal clearance according to the "calculation method" I was advised to use. Few pictures attached. For making this job, I prepared a dedicated driving ring, adjustable parallel, gauge blocks (down to 1um steps) and a micrometer with 1um resolution (and similar error). What I don't have are tools to measure the external diameter of the outer ring and the headstock bore internal diameter with 1um accuracy. With this method I came across multiple issues which make my confidence quite low. I also prepared Excel to support the calculations. IMO the NSK and SKF procedures are quite similar, with NSK being more detailed and well explained. Here are the issues I encountered:

It is very difficult to drive the inner ring using the nut. We used ~ 1.5 meters (5') extension on the spanner to be able to drive the ring up the taper. Appropriate counter bar had to be used within one of the 19mm camlock locker bores. It is much better to prepare the setup for hydraulic press. According to the NSK procedure, the inner ring has to be pushed to some reference point. They recommended to push it until the radial free play/internal clearance reaches 5um. With the bearing put on the taper and very lightly pushed onto place, we started with about 20um clearance and were barely able to drive it up to reach something between 5-10um clearance.

The free play / internal clearance measurement is quite tricky and has low repeatability. Even with light force only, the reading largely depends on the amount of force you apply when moving the bearing up/down. At the beginning of the process, the free play could be felt by hand, but later it could only be observed on the dial gauge. In addition, different readings are received when the outer race is rotated to different positions. I carefully say we reached somewhere between 5-10um, but my confidence in these number is not very high.

Measuring the space between the inner ring and the abutment is also tricky and I could not reach repeatability better than 5um. With the adjustable parallel, it seems that it’s intrinsic parallelism is not in the range of microns, so the reading depends on how you position it in the gap. In addition, how tightly you spread it within the gap. With the gauge blocks, I had to stack up 3 block: 14mm + 1.03mm + 0.991 to 1mm in 1um steps. First, the 1um steps block that would fit depends how much you force it into the gap, and few microns difference do not feel much different... Second, playing around with x3 block takes time and everything gets heated by the touch of the hands. To my help comes the fact that due to the formula, the effect of microns on the internal clearance is reduced my order of magnitude due to the factor, so this specific measurement is less sensitive. Moreover, measuring in different positions along the circumference yield different results in the range of +/- 3 microns (according to the instructions the individual readings should not differ more than 3um… OK).

The most difficult issue is the fact that I don’t have proper tools to measure the bearing outer race and the headstock inner bore with accuracy of 1um. Such internal/external micrometers are super expansive and it seems that those which suitable for 110mm diameter do not reach such accuracy anyhow. The big issue is that from the formula the shrinkage of the outer ring impact on the internal clearance is quite high (about 70% of the interference fit delta).

Even with conservative values that I put into the formula, the difference between the existing ring/spacer width and the calculated ring is quite high- 0.17-0.18mm (check the table below).

Meaning, even if I take no interference fit into account (i.e.0 shrinkage) the calculated spacer is quite different from the existing one. I can't say if this is normal or not, but unless there is a big variance between the old bearing and the “new” bearing, I’m not sure it makes sense.

To tackle these issues, I though of a different approach- Use the formula and the current measures, take some assumptions regarding the interference fit and prepare a space ring that will put the internal clearance in ball park of 7um positive internal clearance. Then semi assemble the spindle and push it into the headstock. Measure the installed/residual radial free play and then use Schaublin’s formula to calculate the final spacer (distance ring). What do you think?

Any ideas are welcome.



  • 1.jpeg
    588.6 KB · Views: 8
  • 2.jpeg
    424.5 KB · Views: 8
  • internal clearance calc.jpg
    internal clearance calc.jpg
    130 KB · Views: 8
Ya this entire process wont work if you cant measure any one of the dimensions required. And measuring anything less accurately than a micron at most will result in incorrect data input in the equation, rendering it useless.

I mean no offense, but this is why companies exist that do this, even without the expertise, the tooling to do one spindle correctly will far exceed the cost of having it done professionally.

I dont think the options you suggested are viable as you cant really stand on the measurements you already took. The free play youre seeing is 5-10 ish microns well your target mounted preload is say within a 2 micron window, and without knowing exact what that play is, you are shooting in the dark. Thats just one of the challenges in your system currently. Without addressing these measuring issues im not sure how to help.
Thanks and no offence :). I talked to multiple local spindle "experts" and I was not impressed that they have better tools nor procedures. I even talked to multiple SKF and NSK distributors and they never heard of anyone who purchased a GN gauge in Israel. They do have much more experience than I do (which I don't underestimate). If it was a viable option, I would bring with me or send the spindle to be assembled somewhere in Europe. However considering the weight of the headstock that must me part of the process it makes this way beyond my economic feasibility.

My assumption/hope is that with few trials with test rings (one if I'm lucky), Schaublin's procedure which depends only on a two measurements- the mounted radial free play and the current distance ring width (which are relatively simple to measure and I have the tools) I can overcome this job.

Another approach I had in mind is mounting the bearing without any distance ring, semi assemble the spindle, mount it in the headstock and measure the radial free play. In an iterative process, take out the spindle and carefully drive the bearing up until I reach around zero free play (let's say not more than 2 micron free play).
On the one hand, I have the concern that driving the semi-assembled spindle w/o a distance ring into the housing will move the bearing further up the taper. On the other hand, the force I had to invest to drive the bearing up the taper is much more than the force I had to apply to take the spindle out of the housing. So I'm not sure if this way is fail-safe. In essence, it's similar to the spacers trial approach, but more flexible.
While lacking any other viable option, I have to come up with something even in the cost of having the need to replace the bearing due to premature failure in the future.
Last edited:
Rock/hard place, I understand. I dont have a real good suggestion, but its worth mentioning, removing and installing the roller can damage it, and the same is true with the ball bearings.

Thinking about being in your position, I guess what I would do is build the spindle without the spacer behind the roller, then use the nut to set the roller to what you "think" is the right preload. Then take a measurement from the end of the shaft to the face of the roller inner race. Disassemble, measure the width of the roller inner race, add it to your measurement, subtract that from the length to the shoulder, and make your spacer that length. Its not foolproof and again may hurt the bearing, but its the only thing I can think of.
Yup I’m aware of the possibility of damaging the bearing (or the spindle in extreme case). It’s a risk I have to manage. Considering that this is not new in box bearing and I paid nothing for it, I can live with that.

I was not able to follow your suggestion. Which end do you mean and which of the roller faces? What do you gain with this measurement compared to direct measurement of the space between the bearing face and the shoulder at some reference point?

Another question- why you don’t prefer the mounted clearance measurement method? I tend to think it’s less prone to measurement errors, which seems to be the big issue here (w/o underestimating assembly/disassembly).
I upload again an older picture so you can see the full spindle.


  • IMG_7525.jpeg
    3.8 MB · Views: 4
Ahh forget my suggestion, I was looking at the print you posted which looks like the roller is at one end, not behind a shoulder.

Without a gn gauge, I do prefer the method recommended by nsk. But in your case you are unable to accurately measure;
Actual housing bore size
Radial movement of outer race
Width to shoulder

So I dont see how you can really do this and achieve a known target radial clearance.
Maybe Im not understanding what youre proposing.
I'll try to better explain myself-
w/o special tools it's not possible to accurately measure the bearing outer race and the headstock inner bore, hence there is no way I can calculate the contribution of the interference fit to the preload. This means I have only one option left which is measuring the mounted free play. This method obviously can only work if the mounted internal clearance is positive. So I aim to positive enough clearance prior to mounting in the headstock. Enough means I'm sure it will not result in zero free play after mounting it into the headstock. Then proceed with Schaublin's procedure which requires only two measurements that I can make:
1. Mounted free radial play - done with the spindle assembled into the headstock.
2. Distance ring thickness at this reference point.

The entire procedure would be as follows:
1. With the spindle on the bench, push up roller until I measure 10 microns free play.
2. Measure the space between the roller inner ring front face to the spindle shoulder / abutment.
3. Prepare a reference spacer, pull out the roller, install the spacer and push again the roller up to the spacer.
4. Semi assemble the spindle into the headstock and measure the mounted free play
5. Using Schaublin's calculation, with the know reference spacer + the measured mounted free play, calculate the final spacer width.
6. Prepare a new spacer, assemble everything back and validate the mounted clearance is in the range of 0-1 micron.

Ok, I follow.

I would add between steps 3 and 4--check free play again to confirm it remained the 10 microns

I thought you were having trouble accurately measuring free play?

Also where did the final mounted clearance target of 0-1 micron come from?