Hari;

I gather from your question you are only looking for a simple formula... as is the case with any CMM, the calibration is done in phases per B89... there is repeatability, volumetric accuracy, bi-directional lengths, single point probing analysis, etc... These results are typically stated in microns, and the volumetric performance (the ability to repeat a value between 2 spheres over the measuring cube's volume), shows the greatest variability.
My recommendation would be to review actual B89 calibration performance in lieu of sales propaganda, taking the volumetric performance, adding to it the bi-directional error (measurement of a known standard such as a gage block), and also adding the maximum repeatability, this sum should be multiplied by 2.
The writer suggests this will give you the "worst case" spherical error of measurement of the CMM as a gage.

Your Partner in Quality

Alvin

http://www.qualitytechnologies.com