[mackeym]

For lathe spindle bearings, i know one good option is to use two angular contact bearings on the far left, and a plain bronze taper bearing on the right. I was thinking about spreading the angular contact bearings apart to both ends of the headstock (while keeping a back-to-back arrangement and preload) and eliminating the bronze bearing. Is there any obvious reason why this would be problematic or inaccurate?

Bearings: ABEC 1 with 40 degree contact angle


For the spindle, i'm thinking about making it out of mild 1080 steel. If a plain bronze taper bearing is used, is 1080 steel compatible in terms of wear with the bronze bearing (assuming the spindle is polished to an optical finish with 0.3um and 0.05um polishing powder). Does anyone know what spindles are typically made out of (hardened and ground steel?)?

[HuFlungDung]

Heat buildup in the spindle lengthens it. The longer the distance between bearings, the greater the accumulative effect of heat, which then results in a change in preload.

I wouldn't use a plain bushing in a modern machine. Rather, a cylindrical roller bearing used at the outboard end can provide good centering without requiring any lengthwise preload consideration.

ABEC1....do you have to look hard for those?

[RICHARD ZASTROW]

mackeym, I've seen near every kind of bearing used on lathe spindles. All have limitions and advantages.

If you use angular contact bearings, keep in mind the spindle shaft will grow with heat. The farther apart, the greater the growth and increase in preload. Also, an ABEC 1 is not the most accurate, it's the least. I'd rather see an ABEC 5 or better.

You will get many recommendations on material. Personally, I like 4140/50 or 4340. Your spindle should be hardened to improve its mechanical properties.

If you elect to use bronze bushing type bearings you MUST harden and grind the bearing surface of the spindle or you will wear the spindle as well as the bushing; much more hassle to replace the spindle than the bearing.

You might also consider tapered roller bearings (Timken brand has become almost a generic description). They can be had at the low-cost-not-so-accurate level as well as the high accuracy / expensive end.

[NC Cams]

If you choose bearings by TYPE and not true calculated bearing capacity versus applied load potential, you have no clue or assurance that the bearings will live let alone perform. Somewhere on the 'Zone, one of my prior rants about bearing load life calculations will go into further detail why this is the case.

The A/C's are there to control axial displacement and it is far away from the spindle for a reason - low RADIAL capacity. The bushed bearing is/was placed next to the chuck to accept the high radial loads inherent to lathe loading at that position.

Don't want the friction of a bushing? Consider a precision cyllindrical, and specifically one with a tapered bore. These are adjustable so as to set the internal clearance. THe use of fixed clearance garden variety cyllindricals have too much internal clearance for good finish and/or tolerance control.

You can run properly opposing tapers or A/C's, providing you account for axial growth under heat so as to NOT overload the bearings if the spindle grows with heat or unload the preload if the mounting position causes the opposite situation.

IF however, you chose to take a SWAG and run what you have giving no concern about load/life calculations, you can pretty much use anything. It may or may not work and nobody who REALLY knows ANYTHING about bearings will be able to say if it will work or how long it will live.

Caveat emptor.

[mackeym]

ok. Thanks for the excellent tips and information. I'm going to go with 2 angular contact bearings (light preload) on the left side and one or two tapered roller bearings (or 2 angular contacts) on the right....and also think about thermal expansion some more and do some calculations.

One more question. To eliminate radial play between: 1. inner races and the spindle and 2. the outer races and headstock, I was thinking about making the parts a little bit too small/big, heating them up, push them together, so they lock together when they cool down. I would heat the parts to acceptable temperatures (not too hot), and also do some calcs to see what kind of expansion to expect. One fitting would be left loose to allow the spindle to expand axially. Are there any obvious problems or concerns with this?

[Glacern]

Interference fit (press fit) can be a real PITA. I much prefer a modular means of locking the bearings in place... usually a cover plate for the outer race and shaft collar nuts for the inners. I preload using wavy disc springs.

I'm with DZASTR on the spindle material. Among those he listed, 4140 Prehard is my material of choice.

[mackeym]

Is it possible to turn 4140 prehard or do you have to machine the unhardened metal and then have it hardened? If it's hardened after machining, will there be much dimensional change from the hardening process?

[HuFlungDung]

By the questions you are asking, you're a long ways from ripe yet Pretty ambitious project for a beginner

4140 prehardened is machinable with carbides quite readily. Drilling it for great distances is a bit more difficult due to the casual user not having the best equipment available for the job, but if you take your time, and use lots of coolant, you can drill with with HSS drills.

The bearings are usually a slight shrink fit on the spindle, and a very close 'tap in' fit in the housing. This is due to the simple fact that assembling/disassembling the contraption is a beast of a job if you do it otherwise.

If you go with two sets of angular contact or tapered rollers, one set should be constrained within a housing, but the other end should be able to float (axially lengthwise) otherwise you will still have undesirable preloading occurring due to differential heat expansion between the spindle and the housing. The spindle runs hotter than the housing in most cases.

A single tapered roller or angular contact bearing is useless. You would be better off with plain ordinary radial ball bearings on the outboard end of the spindle. If you have the bucks to spend, you can find ball bearings with lower than normal clearance, even precision grades might be available. However, the outboard bearing is just a steady basically, the front bearings do most of the work.

One question: how are you boring your headstock? Are you quite certain that you should not be starting off with the purchase of a lathe, new or used?

[RICHARD ZASTROW]

mackeym, None of us here who are attempting to help you have enough information about your project. Give up more specifics and we can be more helpful. When it comes to bearings, pay attention to NC Cams, he's more up on the subject than most or all of the rest of us. He does this stuff 'cause he's dedicated. I do it 'cause I'm retired and bored stiff.

[mackeym]

Thanks for more good info. Yes, i did think about redoing a used lathe. I was going to retrofit a micromark mini lathe. However, the only thing that would be left is a headstock, so i decided on building a larger one from scratch. (i'm going to use SHS25 rails for both axis's and ballscrews with AC brushless servos). I already have the ballscrews (C5 and better preloaded), one set of SHS (the other on the way), i have the motors and drives setup and working, and two 10:1 gear reducers (Bayside Stealth model).

For the headstock, i wanted to use a 12" x 12" x 2" granite surface plate for the base (i know that a special drill bit is needed for drilling mounting holes). I would then mount two 5"x5"x5" angle plates onto the surface plate giving two vertical supports. Then buy 1" x 8" x 18" flat ground stock (low carbon) and bolt this vertically onto the angle plates. But, before the ground stock is mounted, i will mill (CNC) out an area for the bearing outer races. I think that alignment of the holes for the outer races is important, so i plan on coming up with a way to ensure the alignment (after they're mounted) is better than 1mil.

[HuFlungDung]

And, is this going to be some sort of ultraspeed machine?

[mackeym]

haha, no I'm more interested in accuracy than speed (trying to get 1mil or better).

[NC Cams]

The A/C's of a P5 (ABEC 5) tolerance should be fit with a shaft tolerance of "h4" for ID's from 10 to 80mm and a "js4" for 80 to 200 mm shafts. For a P4 (ABEC 7), use "h3" and "js5", respectively.

THe same fits would suffice for tapers BUT finding ABEC 5 or 7's will NOT be easy or cheap. You might want to loosen up the tapered fit a bit to account for the sloppier tolerances you may have to live with in tapers.

The above yields a target INTERFERENCE that varies from 0 to 2 and up to 5 microns according to the size of the shaft and bearing ID.

The Fits in housings depend on whether the end has to remaing free or floats.

In your case, you could FIX the A/C end and, if using tapered rollers, set them up with as close to 0.0000" end play as possible and either DB or DF mounted but then let them float. This would eliminate any thermal growth issues from affecting the axial load on the tapers.

For a fixed end bearing (the A/C's in this case), the housings for P5 grade bearings get fit with a "JS4" while P4's get fit with a "JS3"

For a slip fit bearnig (the tapers in this case), the housings should be fit at a H4 for P5's and H3 for P4's.

These housing fits range from 0 to 5 to 10 microns LOOSE.

Any bearing handbook will explain what these metric fit tolerances mean in numeric terms.

Anything short of the above tolerances would constitute SWAG fits which may or may not fit nor work. You're quite welcome to follow or ignore the above industry standard machine tool fit recommendations. One makes their choices and takes their chances.

EDIT: For those interested in my prior rants regarding the critical need for proper bearing fitment, do a "bearing fit" search on the website. Basically, if you want the optimum performance and accuracy from a device, especially a machine tool spindle, bearing fits are of paramount importance and should not be taken lightly. END EDIT

RE-EDIT: Shaft materials can vary but need to be properly selected in order for the shaft to live/perform properly. For optimum stability and accuracy, the shaft should be turned and then heat treated and then finish ground. For rolling element bearings, PHT4140 would work if it were H&T to Hrc 28-32 minimum. I'd like it better if it were H&T to Hrc 44-48 as this would make for a much tougher, stronger part.

An alternate material for sleeve bearing/bushed shafts would be a carb and hardned steel. A garden variety yet good steel to use would be 8620. For a small price premium, get 8620 AQBQ (aircraft quality/bearing quality). The AQBQ grade is cleaner and nicer to machine.

The problem with any procured steel anymore is that mill run steel seems to be all over the place in hardness and residual stress. Hence, we find that billet tends to move around quite a bit, especially when you cut into it. Shafting is especially sensitive as it twists and bends as you start to machine it.

The fix: leave grind stock on the part. Subject the part to a 1200 deg stress relieve for 1-1.5 hours minimum prior to subsequent H/T. Cool to room temp prior to H and T.

4140/4150 is H and T to Hrc 44-48, double draw if possible.

8620 then gets 0.060" case carburized for use with sleeve bearings/bushings. Harden and double draw.

With either steel, it is always good to cryogenically stabilize and stress relieve after cryo.

Do the above and you're shaft will be darn near bulletproof, stay straight as an arrow and probably outlive you as long as you don't do something really dumb when you design/machine it.

BTW, we do custom/specialty OD and ID shaft grinding of round parts for specialy applications in case you can't find someone locally to do it. Send a paper drawing for a quote.

END RE-EDIT

[mackeym]

Thanks again for the info. I'm working on figuring this stuff out and finding suitable bearings. l'll post questions, comments, ideas, new headstock plans, as they come.

[mackeym]

For bearings, i found 2 options:

1.
Taper roller bearing FAG 32016X.P5 Cost ~$150 ea.
This is claimed to be a "Iso Class 5 (RBEC 5) tolerance" bearing. A P5 bearing. I would guess that this has an accuracy equivalent to an ABEC 5 bearing.(?)


2.
Angular Contact bearing SKF 7216 BEGAM Cost ~$150 ea.
This is an "explorer" bearing and has a dimensional accuracy of P6 and running accuracy of P5. These have the advantage of simple preloading: the faces are ground so that it is preloaded by just clamping two inner or outer races together.

SKF has a nice online load-life calculation. The angular contact bearings can handle a decent amount of weight. 1000lbs in both radial and axial directions at 1000rpm in iso680 oil will last 68yrs (worst case senario) of continuous running. This load-life should be fine for anything i'll be doing.


Comparison of each:
Advantages of the A/C: Easy preloading
Advantages of the taper roller: Higher load capacity. Higher dimensional tolerance/accurcy.
Spec common to both: Running accuracy of P5



Right now i'm leaning towards using two pairs of angular contact bearings because the preloading is easy and i don't need the higher load capacity or dimensional accuracy of the tapers.


Regarding Shaft and Housing Fits: Basically, to get a reliable fit and follow industry standards, the shaft must be ~2 and up to 5 microns larger than the bearing inner diameter for both bearing pairs. The housing fit for one pair should be 0 to 5 (and up to 10) microns loose. The housing fit for the other pair should have ~2 to 5 microns of interferance. I think it may be good to have the housing interference fit on the right side (adjacent to chuck) so thermal expansion doesn't affect axial accuracy. and the loose housing fit on the left.


For shaft materials, i think i'm going to go with 4140 PHT and possibly have it rehardened to higher hardnesses. Not yet sure how i'm going to bore it (or get the housing machined to the required accuracy). I'll think about those things next.

I may post a picture of the proposed design sometime this week.

[gotis]

For bearings, i found 2 options:

1.
Taper roller bearing FAG 32016X.P5 Cost ~$150 ea.
This is claimed to be a "Iso Class 5 (RBEC 5) tolerance" bearing. A P5 bearing. I would guess that this has an accuracy equivalent to an ABEC 5 bearing.(?)


2.
Angular Contact bearing SKF 7216 BEGAM Cost ~$150 ea.
This is an "explorer" bearing and has a dimensional accuracy of P6 and running accuracy of P5. These have the advantage of simple preloading: the faces are ground so that it is preloaded by just clamping two inner or outer races together.

SKF has a nice online load-life calculation. The angular contact bearings can handle a decent amount of weight. 1000lbs in both radial and axial directions at 1000rpm in iso680 oil will last 68yrs (worst case senario) of continuous running. This load-life should be fine for anything i'll be doing.


Comparison of each:
Advantages of the A/C: Easy preloading
Advantages of the taper roller: Higher load capacity. Higher dimensional tolerance/accurcy.
Spec common to both: Running accuracy of P5



Right now i'm leaning towards using two pairs of angular contact bearings because the preloading is easy and i don't need the higher load capacity or dimensional accuracy of the tapers.


Regarding Shaft and Housing Fits: Basically, to get a reliable fit and follow industry standards, the shaft must be ~2 and up to 5 microns larger than the bearing inner diameter for both bearing pairs. The housing fit for one pair should be 0 to 5 (and up to 10) microns loose. The housing fit for the other pair should have ~2 to 5 microns of interferance. I think it may be good to have the housing interference fit on the right side (adjacent to chuck) so thermal expansion doesn't affect axial accuracy. and the loose housing fit on the left.


For shaft materials, i think i'm going to go with 4140 PHT and possibly have it rehardened to higher hardnesses. Not yet sure how i'm going to bore it (or get the housing machined to the required accuracy). I'll think about those things next.

I may post a picture of the proposed design sometime this week.

Well, Don´t know what size you are looking for but I have 2 pairs of 7013 and 2 pairs of 7012 superprecisionbearings 4 sale.

FYI: these are 1350$/pair here.

I want 200$ /pair.

http://www.emersonbearing.com/SpecSh...erPrecison.pdf

[NC Cams]

News flash: ABEC (ball) and RBEC (roller bearing) accuracy specs are for 2 entirely different types of bearings. Until/unless you compare the specs, line by line, spec by sped, you should not assume that the absolute magnitude of each class is/are are equivalent. Details why a bit later

The most important spec you want/need to worry about is concentricity of the inner ring rotation, one to the other and/or radial runout of the rotating ring. THis is why eccentricity highpoints have to be marked and matched when mounting precision bearings in duplex fashion.

I know of a particularly embarrasing situation where two bearings of different sizes but same accuracy class and differeing types (roller and ball) were coaxially mounted on a shaft. This was done to gain the relative merits of each with regard to axial and radial capacity in a very tight package space.

Imagine the surprise on the engineer and customer's faces when after a few revolutions by hand and under NO load, the shaft fitted with this concoction would sieze up totatlly. When they took the thing apart, NOTHING out of the ordinary was noted, no scuffing, no scoring, NOTHING. There was absolutely NO damage.

So they reinstalled the bearings and started rotating it again - they did this a number of times in fact. Same thing each and every time: somewhere between 8 and 12 revs after you started rotating the shaft, the thing would sieze.

The problem was a mystery until you looked at radial runout of inner rings between the two bearings. Although they were of the same accuracy ranged, the bearing SIZES allowed them to have different MAGNITUDES of radial runouts.

Result: after a number of revolutions, the axis of rotation estabished by the one bearing "shifted" due to allowable runout. Whe the same thing happened to the axis of revolution of the other bearing - siezure..

Basically, after a number of revolutions, the "axis shifts" were to the point that they were so much in conflict that the internal clearance was consumed and the shaft siezed - it had no choice but to sieze.

Nothing and I do mean NOTHING but absolutely perfectly matched and assembled bearings would fix the situation. Since this was a high volume automotive application, where speed of assembly is critical and hand selection/fits all but impossible, the design was shipped "back to the drawing board for further study".

Don't forget, by the way, about housing bore concentricity to each/one another.

The higher the bearing accuracy the closer the housings on either end need to be with regard to coaxiality. If they are not, the shaft will have a tendency to sieze as the shaft axis orbits about as the ball and/or roller runouts combine with the raceway runouts.

BTW, the coaxiality factor becomes more and more critical as you both increase the accuracy of the bearings AND as you preload them. The higher accuracy bearings have less self centering/clearancing "slop potential" and the preload takes out the radial and/or axial clearance. THus, the bearings are even less tolerant of "off" conditions.

WHen you are dealing with fits in the single digit mircron level and runouts at or near the same level, you really have to dot your I's and cross your T's when you start to size and fit bearings.

NOTE RE: "capacity" = just because the bearing has 1000 lb load CAPACITY the does not mean you can apply that much load. Most fatigue life calcs ultimately assume that only a small fraction of the rated capacity will or can be applied for a 100% duty cycle. Apply the rated load, even for a short time, and the bearing life can/will drop in a precipitious fashion.

Again, I can cite stories where an engineer did a "catalog load rating review" life calc, selected the bearing and the client used it. Imagine the surprise when the bearing failed almost immediately.

Just like in the useage of accuracy specs, load life calcs need to be done properly and likewise interpretted in order to realize the full life and performance potential of any bearing.

[mackeym]

gotis,

Do you have any similar but larger bearings? 16's (80mm ID)

[gotis]

gotis,

Do you have any similar but larger bearings? 16's (80mm ID)

No, sorry i don´t, these where for a cnc lathe project I was working on but I found a complete spindle cartridge instead.

By the way, these are paired bearings, I can only sell them in pairs

[RICHARD ZASTROW]

mackeym, My last spindle design used a pair of A/C's similar to your proposed at the chuck end. (use good seals) and a floating deep groove radial and nut just ahead of a belt drive pulley. The rear cover had another radial bearing in it help support the tail end of the shaft against belt tension. (Sort of a belt/suspenders arrangement) Of course, you must still observe the basics (see NC's remarks). I don't think I would hold a second pair of A/C's on the right end. If you want the extra capacity, why not mount them as tandem duplexed pairs behind the chuck?