Hallo Jono

A few thoughts for you to consider.

Just too look at it from a slightly different viewpoint.Whats sufficient precision for what you want to do?How little precision is enough?The figures you quote(1-1+1/2 thou in) are IMO flat enough for the bearing to tolerate.Provided you meet the requirements for precision over a total length,which is the most difficult,in measuring-for a start.Also its the "frequency" of the variation,you want a nice slow variation so lots of contact with your rail(spot/sq area).You can look up standard mills precision,its maximum probably 1/1000in per 12in..You have probably observed like me that many are bolting there rails straight to RHS,with no apparent ill effect.This suggests the lower limit for precision without shortening bearing life is substantially higher than we may think.


The limiting factor in how unflat a surface you can use a bearing on is bearing life which is determined by the maximum individual ball loading .If exceed failure occurs through the classic bond breaking,metal flaking equals pitting mechanism.They do quote a figure in hiwin catalog from memory.Calculating is possible for experts?It will be affected by rail bowing(shifting load to fewer balls),preload(consuming some of load carrying capacity-so thats a point against using preloaded rails),acceleration deceleration rates(weight transfer)working conditions(vibration,chatter),load.Whats the worst case scenario?

You need also to consider whether in fact the "static"precision achieved is not swamped under loaded "dynamic" conditions by buckling in the rail support or structure.You could test by creating a jig and measure deflections under realistic loads using DTI.The "buckling" stiffness really determines how worthwhile a higher precision/stiffer(wider,preloaded) rail is.

Because of the substantial load carried by the rails this does provide some of the benefits of "preloading"to the x and y axis(but not the z)Especially on the longest axis.
Medium preload=1/3maximumload,newtons/10=kg). Would this be enough?Preload is applied to all bearings all the time of course,whereas gravity doesnt achieve this.


Larger rails may be stiff enough to pull the mounting to there flatness rather than the desired reverse,and more difficult to pull straight.They will not be true on arrival.


Have you considered how temperature variation in work area will affect stability of dimensions?