Stupid question. If you’re using Easel to cut the square, is it set to cut outside the line instead of on the line?
Another stupid question: You’re not using a ruler that measures in 1/10 increments of an inch and mistaking them for 1/8", are you? Theoretically, it would be easy to use one of those stupid things and assume that something like 10.3 inches is actually 10-3/8", which is .075" off. Of course, someone like me would never make a mistake like that in front of coworkers. Twice.
What length ruler/scale are you using to calibrate? It’s best to try to use the longest and most detailed scale that you can find, like a 600mm scale. If you’re using a small one, like 6", than any error due to readability or other reasons would double at twice the calibration distance. So using a 6" scale and being off by .020" could add up to a .100" error across a 1000x1000 machine.
If you want to check whether the error is due to runout/wobble, try to cut a 3" line that’s .050" deep (in Easel you can just make a square that’s .001" tall and cut on the line). Do the cut, then measure the line width with your calipers. If it’s effectively the same as your bit diameter, then it’s not runout.
If you’ve checked the easy stuff like V-wheels, pulley screws (use loc-tite!), and belt tension, then you may want to make a more quantitative adjustment to your potentiometers. I really like the idea behind your method, but I’m not sure how well it would relate to the high and low voltage settings.
When I calibrate, I don’t do it under load. I still do it the way I showed in the video I posted in the maintenance video thread (X-Carve Maintenance/Troubleshooting Videos - Add Your Own!), except I now turn the V-bit 90 degrees and try to read across the flat instead of using the point. It helps to cut down on the parallax error.
One word of caution on the calibration, either flatten your wasteboard before calibrating or make sure your bit is high enough that it won’t jam into the board. I almost did that once, and would likely have broken the tip off of the bit. It’s bad enough that I’ve accidentally tried to drive two of them through my aluminum touch plate…