Fixed Bearing w/ Belleville washers

[CNCDevil]

This thread is a question about mounting unmatched A/C bearing in a homemade housing, using Belleville washers to achive a so called "Constant Pressure Preload".
I know that this is just another thread in the line many, but I couldn'd really find alot of posts where this subject is discussed in detail.


In the early stage of building my new router, I decided to make my own fixed bearings units, because they were awfullly expensive. How hard could it be just to mount two bearings in a housing? - Now I have learned why

The nuts and screws, which are the heart to accuracy of the system, won't be to any good, without the use of proper bearings. I just mean, that the quality of the bearing setup, should equal the quality of nut/screw. My learning curve regarding A/C bearings have been quite vertical these days.

A week ago, I bought some NSK/IBU 7201 B A/C bearings from Ebay, to use with my 16mm ball screws. As I was digging into the subject of fixed bearings, I got pretty frustrated. I've found out that it is impossible for me to make a decent DB or DF preload configuration. As the more "educated" person can see, there is not given a lot of details about the bearings, with good reason. They are of course unmatched, leading to the fact that it is difficult to preload them properly in either a DB or DF configuration.

I've wrote NC Cams to give his opinion on just setting some preload on the bearings with a torque wrench, but he told me to buy matched pairs and fabricate exact spacers, otherwise I could easily ruin the bearings, because preloading with a torque wrench was too unpredictable.

Problem is, that I don't think it's possible for me to make those spacers precise enough. Another problem is the money required to buy those matched bearings.

-------------------------------------------

Well, here is the deal (and please correct me if I'm wrong)

When reading on the NSK homepage, I saw this Constant Preload configuration and this seems to be my only way out. As you can see on my drawings, the bearings are seperated with two Belleville springs in series. Combusting the washers with 75% of their span, gives a preload of 1170N. The preload is forced onto the two bearings, given that they will each be dealing with 585N.

The force from the ball screw is around 200-300N, so the preload should be heavy enough to withstand that force in either direction, or?

The nominal heavy preload for that particularly bearing is 78.5N, but I assume that 585N will not be any deal to the bearings? Does that EXTRA HEAVY preload affect the bearings in a negative way? Important is, that the bearings are always preloaded, I guess. This should just be a way to have some kind of control over the situation.

Long thread, I know. Hopefully it can be useful to other members.


CNCDevil

[snoopy27]

Originally Posted by CNCDevil This thread is a question about mounting unmatched A/C bearing in a homemade housing, using Belleville washers to achive a so called "Constant Pressure Preload". I know that this is just another thread in the line many, but I couldn'd really find alot of posts where this subject is discussed in detail. In the early stage of building my new router, I decided to make my own fixed bearings units, because they were awfullly expensive. How hard could it be just to mount two bearings in a housing? - Now I have learned why The nuts and screws, which are the heart to accuracy of the system, won't be to any good, without the use of proper bearings. I just mean, that the quality of the bearing setup, should equal the quality of nut/screw. My learning curve regarding A/C bearings have been quite vertical these days. A week ago, I bought some NSK/IBU 7201 B A/C bearings from Ebay, to use with my 16mm ball screws. As I was digging into the subject of fixed bearings, I got pretty frustrated. I've found out that it is impossible for me to make a decent DB or DF preload configuration. As the more "educated" person can see, there is not given a lot of details about the bearings, with good reason. They are of course unmatched, leading to the fact that it is difficult to preload them properly in either a DB or DF configuration. I've wrote NC Cams to give his opinion on just setting some preload on the bearings with a torque wrench, but he told me to buy matched pairs and fabricate exact spacers, otherwise I could easily ruin the bearings, because preloading with a torque wrench was too unpredictable. Problem is, that I don't think it's possible for me to make those spacers precise enough. Another problem is the money required to buy those matched bearings. ------------------------------------------- Well, here is the deal (and please correct me if I'm wrong) When reading on the NSK homepage, I saw this Constant Preload configuration and this seems to be my only way out. As you can see on my drawings, the bearings are seperated with two Belleville springs in series. Combusting the washers with 75% of their span, gives a preload of 1170N. The preload is forced onto the two bearings, given that they will each be dealing with 585N. The force from the ball screw is around 200-300N, so the preload should be heavy enough to withstand that force in either direction, or? The nominal heavy preload for that particularly bearing is 78.5N, but I assume that 585N will not be any deal to the bearings? Does that EXTRA HEAVY preload affect the bearings in a negative way? Important is, that the bearings are always preloaded, I guess. This should just be a way to have some kind of control over the situation. Long thread, I know. Hopefully it can be useful to other members. CNCDevil

This is interesting to me since I am looking at some bellevile washers to take the backlash out of a small acme srew brass nut, by splitting it. I am no expert on this issue, but have a few question after doing some light research on this issue.

1. Will the washers you are using place even pressure on all sides of the raceways?
2. How are you determining you won't destroy the bearings with excessive preload force? It isn't clear to me from your comments or drawing to me yet?
3. What type of Bellevile washers are you using?
4. Have you considered wave type Bellevile washers since they place even pressure when compressed in all directions? Although they limit surface contact area they might be worth considering.

Would be very interested to know how you make out with this idea.

[CNCDevil]

Hello snoopy27

1. Belleville washers will apply even force around the rings.

2. I just know that the bearings can withstand a lot of force. I don't have the number, but we are talking 5-10 times the preload force. I think the preload force has something to do with the stiffness of the radial and axial movement.

3. The Belleville washers are DIN 2093. D=31.5mm d=12,3 W=1.25mm

4. I haven't considered wave type Bellevile washers. I don't think they can apply as much force as the coned B/W. Besides, I'm pretty sure that coned B/W will apply a more even force around the rings.


I think I will just try to make those bearings and see what happens. The forces applied by the ball screws aren't that big, so I really think that it will work out. I could incorporate some extra space in the housing, so that the force can be ajusted by shims, as needbe.

[RICHARD ZASTROW]

CNCDevil, Get a bearing catalog that has your bearing specs in it. In the technical sections load, speed and bearing life expectations are usually described.

My guess is you will find load will increase with preload, speed and life expectancy will decrease.

I agree with NCCams, use a matched set of bearings at the level of precision you require.

By the time you get a "custom substitution" for a proven method perfected, you'll probably spend more $$$$ on the "substitute" than the proven method anyway.

Dick Z

[CNCDevil]

First of all, you are right. I know that this is the professional way to do it and this is what all high-end machines got, but I have come up with another idea, that I want people speculate about.


As we agree, the preload is set to obtain radial and axial stiffness applied to the spindle. The preload is also important to ensure, that the balls will not tumble between the two raceways. I assume this is why the manufactures also state a "miminum load". Please correct me, if my assumptions are wrong.


In this setup, think of a DB configuration, with the bearings spaced apart e.g. 5mm, not making contact at all:

Let's say that the outer rings of the two bearings are independently fixed in place inside the bearing housing. There are two B/W washers in place outside of the bearings, applying a constant force of 78,5N to the inner rings (nominal heavy preload). This should ultimately cause a differential movement between the outer and inner rings, equal to the space between the inner rings in a true DB configuration, where the inner rings faces make contact, when the lock nut is firmly tightened.

Professional unit: Differential space gives Force (preload)

Unit with B/W: Force (preload) gives differential space

My point is, that it would cause the same effect.


Assumptions/experience:

Thus the lock nut is only fitted by your fingers, it will tightly fit the spindle in place (direct contact of metal to metal). The fitting by fingers will not cause any significant preload to the inner rings. All in all, you now have good control over the preload, knowing that it will be within an acceptable range of what's the nominal values. NC Cams wrote, that the more preload, the better, so a little extra will not hurt. Besides, it's far, far lower than the bearings load rating. All in all, the balls are now unable to tumble and the spindle will be well fixed.



Does this sound too weird? I KNOW, that this is not the professional way to make a unit, but if to use unmatched bearings and if it's all about preload control, wouldn't it be the next best solution to make a fixed unit?


What do you guys think?



CNCDevil

[RICHARD ZASTROW]

CNCDevil, There's another little piece missing here. It's "unloading force". This is the force acting against the preload. If the unloading force is exceeded, the balls will "tumble".

I don't have an NSK catalog with that spec., but my FAG/BARDEN book has the spec. for their bearings. B7201C (15 deg. contact angle) is L=108N, M=422N, H=971N.
The B7201E (25 deg. contact angle) is L=136N, M=576N, H=1319N.

In your post#1, you mention these bearings are to be used in a ball screw application. In post#5 you mention spindle.

For a ball screw I'd recommend the highest angle with heaviest preload that the bearing speed will allow. Normally, screws aren't subjected to the speed and radial forces that a spindle must contend with.

I,d still follow CNC Cams recommendation. lol

Dick Z

[mc-motorsports]

Ballscrew bearings and spindle bearings are two diffrent things. The thumbnails attached above make no sense in a spindle. If your building a spindle, just put the bearings in direct contact with eachother. If you REALLY think a 5mm spacer will help, make a set with a rotary grider, ID/OD grinder or surface grinder, .0002" absolute max perpendicular and parrallel. And no, I can't spell for crap.

[CNCDevil]

Well, they are ball screws. Spindle could either mean ball screw spindle or ACME spindle.

As you say "minimum unloading force" and if I understand you correctly, the preload force should be minimum the force that the spindle is transferring to the bearings? Otherwise it would "loosen up" the preload by exceeding this force and lead to ball tumbling? Then the preload should by chosen from the standpoint, on how much load the spindle will transfer to the bearings. If this is not what you mean, then please explain.


Realistic point of view:

My bearings has a 40 degree contact angle, so they can withstand a lot of force. In my machine setup, the load carried on by the spindle will in a worst case scenario, not exceed more than 300N and the spindle speed is low, not more than a absolute maximum of 1000 rpm on rapid moves. The acc. is fairly low too. Also, it will not run 24 hours per day continously.

For my particular application, I think the bearings are pretty well oversized and are actually to be used in a much more aggressive environment. It's not that I not want a professional machine, but if the bearings has a load safety factor that's far beyond what's normally recommended, then it would be unrealistic to put alot of efford in trying to make it even better. I'm a perfectionist and will not tolerate low quality, but I'm trying to see things a perspective here.



CNCDevil

[CNCDevil]

mc-motorsports, I'm not going to put a 5mm spacer in between. The tumbnails in my first post has nothing to do with post number 5. If you read carefully, this is a different setup.

[Mike Everman]

What you are proposing is perfectly fine to do, within certain limits of precision.
Those limits on precision will have to do with the fact that you have a slip fit on the ID of the bearing so the spring can do its work, and the radial movement that this allows between bearing and shaft. There's also the wobble that this clearance can allow, with only the flatness of the BW set to correct it, or more probably make it happen.
Of course it can be made with extremely little clearance, with some care.
Precision ground duplex pairs depend on/require "transitional fits" which you can read as "light press-fit", which makes spring preloading impossible, but the highest accuracy Possible, as there is no radial clearance.

You will probably be fine, but for me personally, I'd keep looking for a duplex pair ground for light preload and spend my time on making the diameter fits right.

[CNCDevil]

Thank you for your comment on this one.

I know my lathe well and can make the spindle diam 0.01mm smaller than the inside bearing ring diameter. I assume that wouldn't cause any significant wobble with the relatively low rpm?

Maybe I'm just the one to try this out and take weekly notices of the bearing blocks condition. Then we could all gain a learning experience and get the issue settled. As I hope, it could be a relatively cheap way for people to make their own blocks, if they would bother to invest the time in the manual work. Me, I have great pleassure every time I get nearby a lathe or mill

[RICHARD ZASTROW]

CNCDevil, Please do not think we're being overly critical, we're trying to be helpful.

The method you propose has been discussed before and pretty much abandoned. But who knows? You may develop a better mouse trap!!!!

Regarding "spindle"; What you are referring to as a spindle I would call lead screw bearing seat. To me a spindle is either a work spindle like where your lathe chuck is attached or a cutter spindle where an end mill is attached. I suppose it's a local thing. lol

Proceed as you wish. Just for fun, keep track of time/cost and see if you actually meet your objective at what cost.

Good Luck!!

Dick Z