Originally Posted by
LeeWay
Not real sure I am understanding everything correctly, but I'll give it a whirl.
Missing balls is not good, but not totally bad. If it is only a couple, I wouldn't expect much backlash. It will wear quicker I would think. There is some space between balls in linear ways, and I would assume the same about ball screws. I need to open one of my screws to check the ball size. I may want to go with over sized balls in the future.
You should be able to determine the actual pitch of the screws. My first ball screws were 13/64" pitch on a 5/8" screw. A little difficult to figure, but once you get the exact decimal value, it is accurate. My second set of screws were a little easier. 15 mm screws with 10 mm lead. I measured this on the screw itself. Count how many groves you see in an inch or CM. Convert either to a decimal value and you have something to work with. The value for my 13/64" screws @ 10 microsteps turns out to be accurate @ 9846.153846 steps per inch. Odd values like that are fine if that is what the math works out to. My other screws are 5080 steps per inch.
If you are driving your gantry or table with a single screw for a long axis, you may be experiencing flex or skewing/racking. This will surely throw the gantry out of square. I built my router using steel and was therefore able to preload the bearing trucks enough to get rid of racking and square the gantry to the table, but less rigid machines might only be able to get rid of some of the racking. Driving this axis with two screws would be the best way to insure that you won't have trouble with racking and unsquare cuts.
If you are getting perfect circles and they are only off by dimension, then I think you should refine the screw lead calibration. This is assuming your spindle and Z axis is solid with little play or flex.
If you have your axes set for 100 IPM, then when the machine travels rapid from cut to cut, it should run @ 100 IPM. To increase the speed during the cut, this is the feedrate, it should be done in your cam program. It may have a preset feed speed for a particular tool you choose in the Cam, but you should be able to adjust to your liking.
On your screen shot, I like to use constant velocity rather than exact stop. It just seems to work smoother. If you have a lot of intricate corners though, exact stop might be better and give more accurate results. It would be slower though. CV doesn't slow down for direction changes much, but exact stop is just what it says it is.
I liken CV to highway driving or Nascar and ES to city stop and go traffic.