V2XT angular bearing backlash

Long story short. almost 1.5 decades I have been messing with thsi machine. 1st got it running with cnc4pc electronic components ran like crap. Spent the $$$$ on Mach motion closed loop servo system ran better. Got as much of the back lash out and have almost accurate steps per inch. Back 10 yrs ago bought some angular bearings to replace the radial bearings that were on the machine. Probably Northrup Grumman maintenance didn't care about this machine. Also had to make new retainer from aluminum. Now .0012" inaccuracy showing up on 21" of travel of X axis in one direction and .0004 in other. I was also getting like +/-.0002 inch in say .5" movements in some cases as much as .0006 but rarely that much. Maybe once or twice throughout movement of X axis. All going in the same direction. Gonna test things with very low accel. Maybe acceleration rate was making ballscrew/bearings move back and forth in the mount. Was going on say 10"/S/S. Now am wondering if maybe need to place wave washer between inner races on angular bearings say .015" thick and and something say .010" thick in between outer races. I am also thinking maybe that retainer might have a groove worn in it so hence .0012" play in one direction, cause when bearings getting pressed to cast iron mounting box no play when going in that direction, but when pressing against aluminum retainer then there is play. So thinking about making a shim that fits teh lip on that retainer. What you all think? Pictures and diagrams will follow.

Here are my calibration attempts attempt #1 atop and #2 below that

will try again with the slower accel rate to see if that makes a diff. Hence strengthening my theory maybe bearings moving in their pocket and along with it ballscrew.