What's new
What's new

Hardness and size tolerance of balls used in ultra precision ball bearings ?

Milacron

Administrator
Staff member
Joined
Dec 15, 2000
Location
SC, USA
Such as used in a precision toolroom lathe with nominal 1/2" balls ? Is E52100 steel, Rockwell C60 and +-.0001" (one "tenth") good enough...... or +- .00005" (50 millionths) required ?

Amazing the price difference from McMaster Carr.....one can get 50 balls for $7.50 at the "tenth" tolerance but to get to 50 millionth tolerance it cost $8.85 just for 1 ball ! (same alloy, same hardness)

I do have reliable instruments capable of measuring these in tenths (Trimos counterbalanced height gage on surface plate with ultra precision Swiss dial indicator as probe to confirm via moving the ball ever so slightly the highest point of ball) so it would be tempting to order the 50 balls and just select the ones (need 11) that are spot on the same diameter. Thoughts ?
 
Give it up with trying to regrade balls. They were already sorted at the manufacturer, far better than you can imagine. You are not only interested in size, but sphericity as well.

Try Bal-Tec.
 
Give it up with trying to regrade balls. They were already sorted at the manufacturer, far better than you can imagine. You are not only interested in size, but sphericity as well.

Try Bal-Tec.
Yes, I was wondering about sphericity and guessed that perhaps McMaster was including that as part of the tolerance figure....but perhaps not...
 
you want to grade 50 balls to save $90? I understand, slow day. gbent has it.
The theory was I would probably luck out and being in the same "run" of balls the first 11 I chose would be identical anyway....and that would be even better than buying 11 of the tighter tolerance balls and not checking them. Plus a little "fun factor" to satisfy my curiosity just how accurate they typically are.

That was the theory anyway.....
 
OK, Bal-Tec can supply .50000" grade 20 bearings (20 millionths tolerance)... dunno price yet. So now the question becomes was that probably the original size of these balls or might they have been .4998" from the get go ? I've been assuming they wore .0002" since new (1970's) but need to be sure before ordering. (I can't get to the manufacturer name without pulling the outer race from the lathe casting, and rather let that sleeping dog stay put if possible...not to mention the fact that info probably wouldn't help identify the original ball size anyway )
 
When if comes to balls for high precision bearings 50 millionths don't cut it. More like 4 millionths size and sphericity and their finish quality is near perfect. Even then, once a set of balls is selected they are graded in size and placed alternately around the bearing whose races were similarly selected from large batches.

High precision bearings are more a matter of grading and selection from the production parts stream than purpose built from scratch but there is a large element of individual inspection and final detailing and finishing..

You don't buy balls like you describe or make them to order in specific batches, you select as sets from a one ton tote full and proceed from there as you propose. Chances are you can find a few sets of a dozen balls matching within 50 millions dia in a lot of 50 balls graded within 0.0001". The size variation will probably follow a curve. A wiser head may safely contradict me: I'm not current on statistical quality control.
 
When if comes to balls for high precision bearings 50 millionths don;'t cut it. More like 4 millionths size and sphericity and their finish quality is near perfect. Even then, once a set of balls is selected they are graded in size and placed alternately around the bearing.

You don't buy balls like that or make them to order in specific batches, you select as sets from a one ton tote full and proceed from there as you propose.
4 millionths....holy cow.....does it matter this is the rear spindle tube bearing and not the front spindle bearing ? Maybe I'd best look a bit harder for the one that got away....
 
When if comes to balls for high precision bearings 50 millionths don;'t cut it. More like 4 millionths size and sphericity and their finish quality is near perfect. Even then, once a set of balls is selected they are graded in size and placed alternately around the bearing.

You don't buy balls like that or make them to order in specific batches, you select as sets from a one ton tote full and proceed from there as you propose.

I second this motion. I went through Mazak's ball screw shop in Japan a few years back and they had huge bins full of balls marked +.0001 , +.0002 -.0001 -.0002 (assuming metric) ect. Even then I was told they would pick through the presorted balls to make up matched sets for each individual ball screw. And this is just ball screws and backlash we are talking about, I've never been through a bearing plant but I would imagine it's a more critical animal.
 
4 millionths....holy cow.....does it matter this is the rear spindle tube bearing and not the front spindle bearing ? Maybe I'd best look a bit harder for the one that got away....

Save yourself some headache - go with a full complement of new 20 millionths balls for the rear bearing and don't worry about the rest. You've got races with millions of ball revolutions worn into them, worrying too much about replacement balls won't gain you much.
 
Save yourself some headache - go with a full complement of new 20 millionths balls for the rear bearing and don't worry about the rest. You've got races with millions of ball revolutions worn into them, worrying too much about replacement balls won't gain you much.
After Forrest Addy's comment I called Bal-Tec back and they have grade 3 (3 millionths) in stock as well. If not much more expensive would that help or due to the race wear you mention a waste of money really ?

( Note to self- Never handle a precision bearing loose ball inner race in an area where if a ball falls out it can roll almost anywhere in the shop and exit via quantum mechanics into another dimension )
 








 
Back
Top