I would recommend investigating using "Remote Load/Mass" in the "Displacement" mode. When used in this way it allows selectively removing degrees of freedom about a local coordinate system.
You can see an example I posted at the bottom of this thread (in this case I was suggesting it more as a fixture but it could also be used as a bearing)
Thanks for taking the time to reply - I've taken a look at the post, but unfortunately can't open the file as I'm still using 2015.
I'm probably missing something, but doesn't using a remote displacement require you input a displacement in the 'free' direction?
Restricting the translation of the outer race only in the radial directions about the L.C.S. seems to be insufficient; I get a large displacement and incorrect solution. Adding the inner race (which is already constrained to it's housing) removes the large displacement, but over constrains the model as some axial force is taken out by top bearing with the remote displacement. I checked this by running a known 'good' model using bearing connectors and listing the connector force.
I think a key point is that the housings of each bearing cannot be treated as ground, as the surrounding structure is also expected to deform and/or move from it's original position.
If you could upload an annotated screenshot or at least hand sketch, I think then more people might be able to chime in
From your original post:
"What I would really like to do is constrain two surfaces relative to each other" -
you could do this bonding across gap (local contact bonding) or using spring connectors or rigid link. Spring connectors may be a good solution as they would let you input radial and axial stiffness. Small amount of axial stiffness might help solution stability
"doesn't using a remote displacement require you input a displacement in the 'free' direction" -
Not necessarily. If you leave the checkbox cleared for that DOF then it won't be restrained. But you may need to otherwise stabilize the solution somehow as you saw with the large displacement warning.
Maybe it is the type of problem where you might need to "Use Soft Springs" to stabilize solution (option in study properties)
Again a simple sketch would go a long way to help others interpret your problem