In article ,
Joe AutoDrill wrote:

Let me add that one of the greatest challenges is keeping the guide rods
from binding up in situations such as I've mentioned above. I may sell
a 36" wide head with 1" guide rods at the ends... When the head grows a
few thousandths of an inch, the guide rods bind. Roller ball bushings
help, but still disintegrate quickly under such loads.

In optical design, design for temperature independence is called
athermalization. There is a huge literature, part of which could be
relevant in that it involves using materials with different temperature
coefficients of linear expansion to passively adjust critical
dimensions. (The non-relevant part involves picking optical glasses
that have contrary variations in optical properties with temperature.)

In a linear array of axes, one would use aluminum and steel in
mechanical combination such that the inter-axis spacing doesn't change
even as the temperature of the assembly changes. One way to do this is
to have a steel frame with the spindle bearings attached to one long
wall by flexures, pushed via a second set of flexures from two aluminum
plates attached to the two short walls and parallel to the other long
wall. If one dimensions things correctly, the two expansion effects
will just cancel.

Joe Gwinn