[hanermo]

Very ambitious.
Good luck.
My scratch built VMC has less than 100 N/um,
and it´s a portal, and uses 35 mm rails, and is version 5 over 10 years.
Measured deflection by standing on the spindle.

Granite is about 3x more flexible than steel.

Got the bearing stiffness data, see attached.

So it turns out the 35mm bearings are more than good enough for 173 N/um, according to that paper, number of axis plus connected objects, in my case a multiple of 5, means the bearings need to be 865 N/um

Then I had a eureka moment while I was wondering how to deal with the 6 months needed before the concrete would be stable. I was designing a connection system that would always stay true and if the cement shrunk you would only lose stiffness and not alignment, then you could remove the plates put a thin layer of grout then reattach the plates.

Then I realised that because I have shrunk my original work-area dimensions down to 300 x 400 x 400, I could now afford to get the concrete core cut from a solid piece of granite.

I replaced the concrete core with a granite core in my FEA and the stiffness lept up to 370 N/um.

.