What's new
What's new

Rules of thumb for setting spindle bearing preload

Many many years ago when we purchased a couple of Monarch lathes, they came with a special spanner wrench. It was used for adjusting the spindle.
The Helical geared Timken bearing lathe instructions are as fallows, "If the spindle is to be operated regularly at fast spindle speeds. it will not be necessary to have the spindle adjusted so tightly in it's bearings. However, if the lathe is used on heavy chucking work, the adjusting collar should be kept tight in order to remove any possibility of chatter. In making the adjustment ,use BOTH hands on the wrench and exert considerable pressure-imposing an initial load pressure on the bearings, thereby removing all end play of the spindle bearings. Owing to the fact that there is little or no wear on the Timkin bearings, it will not be necessary to make frequent adjustments on the spindle" Right from the Monarch manual.
That said, we always set a slight pre load on spindle bearings, and check for under load running temp. I am not sure that using both hands on a spanner wrench is the way to go. but who am I to argue.
A buddy purchased those 70 year old lathes from us about 20 years ago and there still working great with the original spindle bearing in both.
 
What's the most recent edition of the "Timken Engineering Journal"? Is 1960 the most recent?

If so I will probably order them as I like paper books. I found a PDF of the "Timken Enginering Manual", but it's only 220 pages. I have not yet looked at what else they have on their site. I suspect that the daunting part will be finding what I want among all the publications by SKF, NSK, Timken, etc.

Just received my Noga NF16003 and am in heaven after fiddling for years with old style multi-joint stands that liked to drop the indicator. Should have bought a Noga a long time ago. Will probably buy another in the next larger size for my Clausing lathe and knee mill.

Now that I have something which I can adjust properly I see that the spindle bore radius shifts a tenth when I reverse direction, but once it shifts the needle doesn't move at all. Just ran it for 20 minutes and rechecked it. Spindle bore is comfortable to the touch and the TIR readings are the same. The motor speed control electronics are next to the spindle and get warm quickly, so I suspect much of the heat is from that and not the bearings.

I then clocked my ebay Indian MT 2 test bar and am measuring an angular misalignment of the spindle and axis of rotation of about 30" of arc comparing TIR at 1" increments along the bar. I can see the sweep of the needle move back and forth, so at least part of the error is the bearing. To put it bluntly, I'm amazed. This was ~$1250 delivered with tax and shipping.

I'd considered buying a used Schaublin many times, but was nervous about buying something like that off ebay without being able to inspect it first. It would be very interesting to know if my G0937 example is typical or a lucky fluke.

I bought my Clausings 25 years ago because I didn't trust Chinese machines. But they've made great strides.

I am not going to post more about my testing as these are "unmentionables" here. I'm going to review Connelly and then run through his tests. I'll be reporting the results on a couple of groups.io lists, projects-in-metal and machine-tool-rebuilding if anyone is interested. Anyone can read them without joining.

Contrary to allegations in a previous post, I am not an engineer. I'm a research scientist. I had planned to get a job as an instrument maker and technician at a university when I retired, but the academic environment is now so corrupt I'm not interested in getting anywhere close to to a university. I got to know a bunch of those guys when I was at university and found them huge fun to talk to and be around. They never knew what they were going to be asked to do and they had to be able to do *anything* as a small group.
 
  • Like
Reactions: PDW
Hah!

I don't have a problem with the question, it's just hard to answer without knowing the type of lathe and type of bearing. What's good for a Clausing with Gamet bearings might not work for a much bigger or smaller lathe.

But...on the subject machine tool, I wouldn't even worry about it much beyond going by feel and perhaps some basic measurements.
 
..........................................
I then clocked my ebay Indian MT 2 test bar and am measuring an angular misalignment of the spindle and axis of rotation of about 30" of arc comparing TIR at 1" increments along the bar. I can see the sweep of the needle move back and forth, so at least part of the error is the bearing.
A little caution on that Indian made test bar. I bought one to try out. Runs dead true end to end set in vee blocks. But after checking runout on a known spindle, I knew it was running dead true, it shows runout out on the end of the test bar. What I determined was the test bar taper was not ground correctly on the last 3/4" on the small end. Get what you paid for. Ken
 
I used to make up wheel bearings .002 to.0025 - lose and never had a wheel bearing go bad after I installed one. the idea was that after a normal warm up the would be near zero. ..and not be overly tight.
Likely a warmed-up lathe with .001 end play would be Ok.
Counting the threads and dividing it in travel to mm or inch to know the amount of turn will give what take up is an asset..perhaps you find the 1/8" turn will give .001 tale up....so you turn 1/16 to take .0005.
 
Last edited:
I haven't worked on tons of lathes, but I seem to recall a heavy wave spring that provided preload and allowed for thermal growth.
 
I completely agree that generally you get what you pay for. However, I've been pretty lucky. I spotted my bar in the spindle and it had good contact all around. The last inch or so at the tailstock end doesn't seem very good, but until I get the tailstock properly aligned it's hard to tell. In the meantime I just don't use the last inch or so. Similarly I found <0.0001" TIR on both a backplate mount ER32 chuck and an MT 2 version of Chinese origin. The collets are another story. The ones I've measured have 0.0005" TIR.

It's going to be really interesting to test my Clausing 4902 with there MT 4 bar I bought. It's a plain bearing machine. I know from my level that the center part of the bed is worn down about 0.003" but I never had a good way to test the straightness of the prism or the headstock alignment.

My first project on it was a bar with a series of collars at 1" spacing which I attempted to turn separately to a tenth. It taught me a hard lesson about thermal expansion and the need for flood coolant. I hope to succeed in a rematch.

My first lathe was a Sears 109 which is a miniature 1880's machine. I still have it for the simple reason it was more trouble to sell than it's worth. But I'm beginning to develop a perverse fascination with trying to make some tight fit parts on it using nothing but spring calipers. The main job I did on it was rebush the leadscrew. That came off well, but I feel as if I cheated by using modern measuring tools.

I enjoy doing old school bench work with files and can readily hit parallel and square on 2" x 4" plates of 16 gauge steel to a thou using spring calipers and a square. For lots of work I can do it with files faster than I can set up the mill. So trying my hand at late 19th century machining seems an interesting way to spend an afternoon. Modern machines were built by people using such machines and methods. I admire skill in any domain and have spent my entire life trying to learn as many as I could. The fly in the ointment though is that they are transient and I have learned the hard way that things I was very proficient at 30 years ago I can no longer do without significant refresher practice.

120 years ago if a ship broke down at sea everyone died if the engineer couldn't fix it. So a marine engineering school exercise back then was to take a round bar and flat plate, file the bar hexagonal and make a hex hole in the plate so that the bar was a close sliding fit in all 6 positions for the length of the bar. I still haven't worked up the courage for that. But maybe some day.

@michiganbuck I followed the same practice on VWs, but I follow the factory manual on my Toyota pickup which uses friction as the spec. After today's exercise I'm likely to play around with a scrap spindle and housing to get a better feel for how the end play and friction methods relate. I suspect that the change is more related to the demands of factory production than anything else.
 
Precision roller bearings certainly are designed to work under preload. How much preload is correct depends on more factors than I care to type, and either way you are unlikely to unearth the correct exact spec for your machine.

In this case, its not worth the effort you are putting in, and frankly you are likely overthinking it. Usually in a spindle this basic the preloads are set by spacers and you cant overload it by tightening the nuts. You certainly can be unloaded though, which it sounds like you still are if you have axial play by changing directions.
 
Tapered roller bearings are to be set with a preload when new. This is actually a compression of the bearing itself as compared to the thickness of the free bearing. As it is near impossible to measure this axial compression, so rolling torque is used to approximate it. The bearing will eventually wear so the axial preload goes to zero. Most of this wear takes place on the large roller ends and the cone shoulder. This is how bearings are designed to be used.
This rolling torque varies with the bearing sizes so the torque must be obtained from the equipment mfg. most likely near the 20inch # range. It’s rolling torque not static torque.

For used bearings, the break in has already happened and using the new bearing preload will overload the bearing. For used bearings ideally, getting them back to zero preload is what is needed. Again you can’t measure it easily so using a 5inch # rolling torque is a good approximation. You are looking for a guarantee that errs on the side of getting to zero preload without overloading the bearing.

Tapered wheel bearings are a different animal as they are very low speed and their size is determined by the spindle. an axle spindle is designed and bearings big enough to fit over it are used. Invariably the bearings are far larger in capacity than the wheel loads would dictate. You can get pretty sloppy with them.
 
The recommendations by eKretz and Tyrone Shoelace worked fine. I get less than a tenth runout at the spindle face and MT taper and no more than perhaps 110 F after 40 minutes running at full speed with no load. Eric's advice on temperature and Tyrone's advice on maintaining a few tenths of endplay seem pretty sound for any bearing system.

I plan to measure rolling torque eventually just so I know for the future.
 
The recommendations by eKretz and Tyrone Shoelace worked fine. I get less than a tenth runout at the spindle face and MT taper and no more than perhaps 110 F after 40 minutes running at full speed with no load. Eric's advice on temperature and Tyrone's advice on maintaining a few tenths of endplay seem pretty sound for any bearing system.

I plan to measure rolling torque eventually just so I know for the future.
The real proof of the pudding is in the components you produce. The surface finish and dimensional accuracy will tell you wether the job is a success story or not.

Regards Tyrone.
 
I'm not expecting much from the current bearings. The tramp as the spindle rotates at slow speed is very obvious on my tenth indicator. I simply wanted to recover from having jammed an MT 2 ER32 collet chuck in the spindle and having to remove the lock nuts to get a gear puller on it.

It's getting ABEC 7 bearings once the spindle and housing have been properly prepped.

I've been contemplating how to construct a gadget that will allow applying the same measured force in both directions using a torque wrench to apply the pressure. The hard part is generalizing it to handle spindles and carriage/table bearings. My Clausing 8325 was in as new condition except for the 35 year old grease in the feedscrew bearings. I replaced them, but did not see the shims until I got it back together and was putting the old bearings away. So I've got horrible backlash as a result.
 
You seems to be connecting runout to preload. The are independent of each other and don’t necessarily prove much. Having axial play means you have negative preload in the thrust bearing which is not unheard of in some very high speed spindles, but certainly not right in this application.

There is a factor missing in this picture you have painted us. Tapered roller bearings used in spindles are only capable of taking load in the radial direction, unless this has cup and cone bearings like a trailer-then it should be taken out back and shot.

But assuming it’s real roller bearings, there is a thrust bearing in the mix somewhere that you aren’t taking into consideration. Either a pair of angular contacts, or a dedicated high angle thrust bearing.

This thrust bearing setup is what controls axial stiffness, and you can have the rollers preloaded perfectly but if the thrust isn’t, endplay…

Alternatively, preload the thrust setup but not the rollers, and there will be no radial rigidity.

At the end of the day you can absolutely measure the mounted bearing preload in a complete spindle assembly to KNOW where you are. It’s only of real value if you have the spindle specs to compare it to, but if you knew the bearing specs and setup you could get closer than anyone could ever disprove.

In your case you loosened a spindle up and can retighten and be in business, when you rebuild this with new bearings you have a whole bunch more mountains to climb in setting each of the bearings to have the right mounted preload so the final result is a good spindle.
 
"There is a factor missing in this picture you have painted us. Tapered roller bearings used in spindles are only capable of taking load in the radial direction.." WTF! Knowledge of basic geometry should make clear why that statement is wrong.

I am *not* connecting TIR to preload except that if the endplay is not "small enough and large enough" there will be a lot of problems with TIR and wear. The latter being the biggest issue.

These bearings don't have tapered rollers, so my nomenclature may not be strictly correct. I used it simply because it's pretty common to use "taper" to distinguish cylindircal roller bearings designed for radial and thrust loads from bearings intended only for radial loads. Tapered rollers simply increase the ratio of thrust to radial load, but without altering the total load capacity.

From where I sit the "mountains" look like K2, Everest and a few others in a long series of climbs. Regrinding the spindle to +0/-0.0002" is not trivial. And rather scary if you've never done it.

Subsequently I've learned that the bearings are such a tight fit on the spindle that finger tightening the spindle nuts would have sufficed. At the time I did not know that. Hence my question. I submit that the recommendations by Eric and Tyrone are valid for *any* bearing which must take a thrust component whether plain or rolling contact.

I make no claim to being a "machinist" but I will claim to be a meticulous and skilled "mechanic". I have a 40 year track record of taking things apart with no documentation as well as things taken apart and not reassembled by the person who took it apart and reassembling them correctly. Even if all I had for information was the wear pattern on the parts and common practice.
 
You are not familiar enough with the terminology you are using to know how wrong you are. It’s ok, unless you do this stuff every day I wouldn’t expect you to be.

What are you grinding?? The journal’s? A .0002” window is a cakewalk. If you’re referencing the taper:tooling surfaces, that’s not good enough.

It very much sounds like you are confusing some bearing types, but if you indeed do have a cup and cone design like you are describing (just not saying), then you have a different fight ahead of you.

I’m completely lost when you compare plain vs rolling contact. Again I get the sense you aren’t familiar with these.

Taking things apart and reassembling is great, but it’s hard to see microns, preloads and compound misalignments by eye. When you dont have the experience to know what to looks for this is an expensive and time consuming endeavor.
 
Lol, ya, cup and cone bearings, did you read it? Is that what’s in your spindle?

I understand all these dynamics, if that’s what’s in your “spindle” I’d be very surprised about how poor the quality is. I hope for your sake it’s not built with those.
 
I get that you are beyond assistance, it was anyone else that reads this that deserves quality information. Good luck with your project.
 








 
Back
Top