Bearing Block design

[devincox]

I am staring on the x axis on my IH Mill and have been looking at designs for the bearing blocks and have some questions with regard to the bearing block's construction.
I have seen 3 different designs with regard to placement of the AC Bearings in the bearing block (driven side of the ball screw)
1. a shoulder built into the bearing block to separate the back to back AC bearings
2. a spacer/washer placed in between the AC bearings to separate the back to back AC Bearings
3. nothing between the AC bearings

I am curious as to which design is better? Or is it 6 of one half dozen of another?


Thanks,
Devin

[SCzEngrgGroup]

Any of them will work, the difference is in ease of design/fabrication, and adjustment. With #2, those "washers" are shims which are used to set the bearing pre-load, so when you install the shaft, you can just tighten up the nut and get to work. With the other two designs, the nut torque determines the bearing preload, so assembly is more critical if you want to avoid backlash (if the nut is too loose) or rapid bearing failure (if the nut is too tight). Personally, I'd go with #2. Bearing shims are available from McMaster-Carr.

Regards,
Ray L.

[davo727]

If you buy a matched pair of preloaded ac bearings they are ground with the offset built in so you just put them face to face and torque the shaft nut and when the center races are clamped together you are all set with correct preload and no slop.

[Mike Nash]

#1 I figured to be pretty difficult to get accurate. That inner step thickness needs to be very uniform.

#3 is only going to work with matched bearings, you won't have any preload otherwise because the inner and outer races are the same width.

I made two spacers on the lathe for my X and Y. It dawned on me when doing the Z axis that the outer race of a cheap standard bearing of the same OD as the angular contact bearings should be a very uniform spacer, so that's what I'm doing. A bit thick maybe, but cheap.

Is anyone else finding that the time to make all of these parts without a CNC machines is very time consuming?

[devincox]

I already have the bearings and they are not a matched pair, so I am going with #2.
Now a few questions:
Which shims do I get? How thick? My bearings are 15mm x 35mm x 11mm.
What torque do I set the nut to?
I am not sure exactly how to know when the proper preload has been achieved.

Thanks,
Devin

[BobWarfield]

I already have the bearings and they are not a matched pair, so I am going with #2.
Now a few questions:
Which shims do I get? How thick? My bearings are 15mm x 35mm x 11mm.
What torque do I set the nut to?
I am not sure exactly how to know when the proper preload has been achieved.

Thanks,
Devin

Knowing when the proper preload is achieved is a real trick as far as I have seen. NC Cams and other bearing experts seem to indicate there is no easy way to do this and you should send the bearings off to a specialist to grind them so they become a matched pair.

It occurs to me that one could build a test cell that would enable you to set a recommended preload and then measure the displacement caused by that preload and thereby determine the required spacer thickness.

I have some notes on how to create such a test fixture on my web site here:

http://www.cnccookbook.com/CCMillBeltDrive.html

I haven't tried it. I'll be interested to see if someone else has a better idea, or whether the suggestion is to just take a wild guess on the spacers and get things approximately right.

Cheers,

BW

[philbur]

I think the technically correct method for angular contact bearings for precision applications is to buy a matched pair.

In my limited experience I have never come across preload by shims for angular contact bearings, I have always assumed that the difference between no preload and to much preload on small angular contact bearings is a few tenths, making shims impractical. This of course may be a wrong assumption.

I think the generally accepted "not correct" but "workable" solution for things like ballscrew thrust bearings is to use a short outer spacer together with a jam nut or some other nut locking arrangement. Personally I prefer a single nut with a radial setscrew locking arrangement, fine adjustment is a bit easier.

Phil

Knowing when the proper preload is achieved is a real trick as far as I have seen. NC Cams and other bearing experts seem to indicate there is no easy way to do this and you should send the bearings off to a specialist to grind them so they become a matched pair.

It occurs to me that one could build a test cell that would enable you to set a recommended preload and then measure the displacement caused by that preload and thereby determine the required spacer thickness.

I have some notes on how to create such a test fixture on my web site here:

http://www.cnccookbook.com/CCMillBeltDrive.html

I haven't tried it. I'll be interested to see if someone else has a better idea, or whether the suggestion is to just take a wild guess on the spacers and get things approximately right.

Cheers,

BW

[BobWarfield]

A preloaded pair is definitely best, but given that option was already ruled out, shims are what's left.

Interestingly, Slocum's Precision Machine book mentions a method similar to my test cell method:

http://tinyurl.com/6pd7uu

I got that off a quick search. He does refer to shims as less desireable but doesn't rule them out.

I think it's a case of, "Relax, it's a hobby not an aerospace/nuclear reactor quality production tool."

One other consideration is that every account I've read says that trying to figure out the torque on a nut as a way of setting preload just doesn't work. Be careful not to over-torque or over-space too or you may damage the bearings, though that seems less likely in this application than a spindle.

Cheers,

BW

[Cruiser]

I have found that the biggest deal with finding or setting a proper pre-load with a mechanical system would be heat and what happens to the components of that system. If this is a cold system which would only be affected by ambient conditions then taking the bearings to zero and then adding .002" pre-load would be considered safe for the bearings. something like a spindle which gets hot and lengthens in use would require more pre-load as when it gets hot the bearings get sloppy and the pre-load helps to compensate. you may find the opposite to be true if it is the housing of the mechanism which expands and causes the bearings to get tighter ! There is a lot of math and rates of expansion of different matr's per their sizes, or mass etc to calculate anything which would be usable, of which very few people have the knowledge or information to work with. In short, your designing so you decide using your best guess then test the system for a period of time and be ready to redo it if necessary ! My guess is that it will not take much to keep the bearings doing the job you expect. And, I have found that bearings are somewhat tolerant of mistakes more so in a low energy system, but as the energy within the system rises so does the risk of destruction rises. This information is borrowed from the Mickey Mouse school of shade tree engineering written by professor Donald himself !

[philbur]

Yes but he seems to recommend against it. Quote:

"Using shims is a sure way to create difficulty in getting the preload just right so that there is no variation in tare torque. The reason is that a difference in preload caused by tens of microns can be too much. Controlling shim thickness to this level can be extremely difficult. ….. If a manufacturer tries to sell you leadscrew support bearings that require you to make measurements, determine the required shim thickness, and then use shims, tell them that you do not want their bearings."

The Tormach uses a pair of angular contact bearings with jam nuts on each leadscrew. It seems to work OK, but as you say it is a bit hit and miss to adjust properly. I guess for amateur use a bit to much preload on the leadscrew support doesn't show-up as a problem until you have a couple of thousand + hours on the machine, which maybe never. So probably shims or jam nuts are both a bit borderline but we get away with it, providing it is done with a bit of care.

Phil

PS: Interesting reference, I just ordered the book, an early Christmas present to myself.

Interestingly, Slocum's Precision Machine book mentions a method similar to my test cell method:

[philbur]

Strange, I can't find that reference anywhere on Amason.com.

You are probably right, but it is nice to understand what's going on. However I think that 0.002" on small angular contact bearings is probably going to kill them stone dead.

Back to back match bearings are good for applications where heat/expansion may be a problem as the dimension exposed to the differential expansion is only one bearing width.

Phil

This information is borrowed from the Mickey Mouse school of shade tree engineering written by professor Donald himself !

[devincox]

Thanks for the excellent information and resources. I think the best way to proceed, for me at least, is to use the bearings I have now (unmatched) and shim. This is just a hobby for me and I would be happy just to get .001 with the machine. I will however, save up for the matched bearings.
How strange would it be to ask for bearings for Christmas? (c:

Devin