Are you planning to make some decent quality ball bearings? How big?

The nominal dimension is sometimes called " Track Curvature" or raceway curvature" or probably a few other things as well. That is one of the many features that are claimed as proprietary since it can have a pretty profound effect on a bearings speed rating, among other things.
General discussion of effects of nominal value here -
https://nhbb.com/reference/ball-roller-bearings/in...

Tolerances for similar features, but not the curvature are covered by ABEC and ISO.

My WAG is something less than 0.0001" for bearings with bore smaller than 3" or so. Good quality Commercial bearings are very nicely made.

310toumad,

What can your fabricator do?

--
JHG

The ball bearing race characteristic you seem to be asking about is called "osculation". Osculation is the relationship between ball diameter (D) and race track radius of curvature in a plane transverse to the rolling direction (r). Osculation is typically defined as a ratio (r/D) or percentage. For example, if you have a 1.000" dia ball and a race track radius of curvature of 0.530", that would be an osculation of 0.530 or 53.0%.

Ball bearings commonly use different osculation for the inner and outer race tracks. This helps provide a better balance of contact fatigue life between the inner/outer race track surfaces, and more optimized ball/race rolling contact conditions for improved efficiency. The osculation used for each race is basically a compromise between efficiency and contact stress levels. A typical good quality ball bearing might use an osculation of 51% to 52% for the inner race and 53% to 54% for the outer race. I don't know of any rolling element bearing standards that specify tolerances for osculation or race track curvature transverse profile/radius. I've had some custom aircraft ball bearings made using an SCD specifying an osculation tolerance of +/-0.50% (ie 52.5% to 53.5%), and the FAI reports showed all osculation values were well within spec. So meeting that osculation tolerance must not have presented a problem for this particular vendor.

From what I've seen, the tolerance of the track radius isn't typically a driving factor in and of itself. It comes from needing to meet other requirements like osculation or axial or radial play. Also, the tolerance on the radius of the track isn't usually the difficult part to meet, the small values of out of round are usually more difficult to meet than track radius geometry.

Aidan McAllister
Metallurgical Engineer

OP said - "tolerances on a radius for ball bearing grooves..."

Now I'm thinking I may have misinterpreted that. Or maybe not.
Another unfortunate example of a few words, and no pictures or drawings leading to circular, fractured discussions.
https://media.giphy.com/media/d3Fz2JdKIxLOS7te/gip...

So here is a tech document with some annotated drawings of features, bearing notations (which the OP may need) and even measurement principles, among which //may// be what the OP was asking about.
Not the value of the tolerances, just the feature.

• http://files.engineering.com/getfile.aspx?folder=f9691a29-a458-4618-8ec5-0b

The OP asked, "Does anyone know if there is a source for the tolerances on a radius for ball bearing grooves?"

I understand what the OP is asking about, but there is no simple answer to the question. First, there is no published standard controlling this particular characteristic (ie the dimensional or form tolerance of the race track surface profile in a plane transverse to the rolling direction), which is probably not the answer the OP was hoping for. As AidanMc noted, the tolerance of the race track surface profile in the transverse plane is not usually critical. What is more critical with regards to the race track surface are characteristics like runout to the related ID/OD and end face surface, circularity and waviness about the axis of rotation, contact angles in the assembled bearing, etc.

One thing that would be helpful for understanding race track surface tolerances is an explanation of the finish grinding process typically used for ball bearing races. The first operation is a precision centerless grind of the ID or OD surface. Then the end faces are double disc ground at the same time, perpendicular to the centerless ground ID or OD. Then the race track surface is form ground using a type of centerless grinding called "shoe grinding". The shoe grinding fixture locates off the precision ground ID/OD/end faces so that the form ground race track surface is precisely aligned around the circumference. The tolerance of the race track transverse profile will be whatever the form grinding tool provides. The type and size of the form grinding tool used will be one factor in deciding what tolerance is appropriate.

As far as inspecting the transverse race track profile, here's a video showing how it's done.

The leading question is WHY? Knowing this might narrow the field to the best answer.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)