https://www.bearing.co.il/img/tech1/PDF/NSK_CAT_E1...

Take a look at the images in the 10.2.2 paragraph. Notice also how the bearings have a support in the corners where the force transmits itself.

A spacer would help with the preload,ut in your case it would need to be placed on the upper part if I'm not wrong.
(Edit- or, you can put a spacer on the lower part that goes around the outside instead of the inside)

The overall length of the inner spacer as compared to the length of the outer spacer determines the bearing preload. Without the inner spacer, the bearing would be severely overloaded when the shaft nut was properly tightened. Without the inner spacer, it would be impossible to tighten the shaft nut to the perfect amount to preload the bearing. But, the length of the inner spacer can be changed in 0.0001" increments to provide the exact pre-load desired.

Johnny Pellin

"in many drawings I can see the presence of a spacer between internal rings."


Understanding the applications for the components shown in each of those drawings would very likely help shed some light.

Machine tool spindle - stiffness is supreme - "rigid preload" using JJPellin's spacers is VERY common. 15° and 25° angular contact perhaps more common than industrial 40° angular contact bearings used on more mundane equipment making two bearings handle radial and axial loads.

In the old day angular contact ball front wheel bearings were common. They look pretty steep.
http://chevy.oldcarmanualproject.com/shop/1955/ima...

But many or most mfrs evolved to the vastly superior ( IMO )taper rollers, only to devolve in the modern era to assembly friendly wheel bearing cartridges. I'm not sure what contact angle may be common in those.