# Printing a non-self-aligning caged deep-groove ball bearing

I've recently designed a non-self-aligning caged deep-groove ball bearing. Now I'd love to get one 3D printed.

However, assembling those can be tricky and I highly doubt it's even plausible to print them. All the components themselves can be printed without a problem, but I'm not sure whether I'll be able to put them all together in the end.

What are my options?

FDM printers are probably out, although it would be great if I can find a way to use those. Would an SLA or perhaps an SLS printer be able to pull it off?

Of course the thing still has to work (move) in the end.

Two aspects of your question can immediately be addressed, one with much less certainty than the other.

If you were to print the indicated parts using PLA on an FDM printer and assembled as much as possible up to say, the outer bearing ring, you might be able to use boiling water to soften the ring with boiling water enough to get it around the balls. It's also possible that the deformation will effectively destroy the usefulness of the part.

On the other item, SLS is virtually perfect for this type of part. The un-sintered nylon powder provides positional support for all of the other parts being sintered. When the print is completed, the powder is blown out with high pressure air, freeing up the individual components as per the original design.

Be advised that there is to be expected some shrinkage of the parts which will affect any segment of the component that has tight or close clearances.

From Shapeways site regarding SLS printed parts (Shapeways)

Executive summary:

maintain a wall thickness of at least 1mm,
accuracy is 0.1mm,
always keep a clearance of 0.6mm
and parts may have a deviation of 0.2mm.


When designing something mechanical that has to be 3D printed or when adding a mechanical feature to your model there are several main considerations.

Also listed in the above link are the considerations of enclosed parts. Each of the balls in the bearing design may or will be printed as a hollow sphere filled with nylon powder. The bearing rings will also be printed as shells, unless appropriate steps are taken to make them solid.

By the same token, if a part need not be solid and can be designed with holes to "drain out" the un-sintered powder, the cost will be reduced accordingly.

• A clearance of 0.6mm with a deviation of 0.2mm is going to get problematic with parts of a small size (and since most bearings aren't that big, the details get small very quickly. I guess I'll have to re-design the plastic bearings in less-conventional ways.
– Mast
Nov 16 '16 at 16:38

How about using a dual nozzle FDM printer with ABS and HIPS?

You can use HIPS to print any support materials or any spacers that you need. The HIPS can then be dissolved using Limonene. Limonene will not do much of anything to ABS.

Similar ball bearings have been printed like this before. I am not aware of any reason this would not work for you. Other types of ball bearings have been printed in FDM with no support material.

• I didn't know such printers did exist, it definitely sounds interesting.
– Mast
Nov 15 '16 at 23:06
• There is a caution regarding using Limonene and ABS, as part of the chemical structure of ABS is affected by Limonene. Users of this method are cautioned to keep the immersion of ABS in Limonene to a minimum. Nov 15 '16 at 23:17
• Well, there are other soluble supports. That option does work but there may be other options besides HIPS that might work alot better with less degredation of the main material. @fred_dot_u How do you feel about PLA and PVA? 3ders.org/articles/… Nov 17 '16 at 2:07
• I've been using PLA with PVA recently, as I have a dual extruder printer. If I can get the PVA to stick to the build plate, the results are good. The bearing design would be a challenge for my slicer, as it would be difficult to designate the internal spaces to have support placed. People print bearings of this sort in place without supports but the results are somewhat crude. SLS would be the only way to get even moderately useful models of this nature. Nov 17 '16 at 10:31