Christopher_C
06-27-2012, 11:02 PM
Hi! I am a student shop proctor working on adapting a shopbot to serve the needs of our college shop. The shop already has ProtoTRAK CNCs for 24/7 student access that are easier to program and run than the shopbot ever will be, so what the shopbot is adding is bed size. It is a student shop that primarily generates professional-grade prototypes for our industry-relations program, called clinic.
It hasn't been very well used in the past because of a few factors. One is a persistent set of bugs which I am addressing in a different forum. One is fixturing workpieces, which is difficult and tedious since we don't have room or budget to add a vacuum hold-down-but I think I have a solution for that (will post it after testing it out).
The real underlying problem is the lack of accuracy. The machine we have just doesn't seem repeatable to the level of precision we need-what's the point of being able to give it G-code from mastercam to 5 decimal places when the machine is only accurate to a tenth of an inch?
I've been using the C3 command, find home using prox switches, and then sending it to a pre-determined part of the table and cutting down, only to find it landing randomly in a circle about .1" in diameter each time (not distinctly spread out in X or Y). That's almost ten times more slop than would be acceptable for the work it needs to do to be widely used; we just don't handle too many jobs that are both large-format and that sloppy. Is that a normal amount of error? Are there any good tricks for cutting that CEP down to the order of a few mils at most?
It hasn't been very well used in the past because of a few factors. One is a persistent set of bugs which I am addressing in a different forum. One is fixturing workpieces, which is difficult and tedious since we don't have room or budget to add a vacuum hold-down-but I think I have a solution for that (will post it after testing it out).
The real underlying problem is the lack of accuracy. The machine we have just doesn't seem repeatable to the level of precision we need-what's the point of being able to give it G-code from mastercam to 5 decimal places when the machine is only accurate to a tenth of an inch?
I've been using the C3 command, find home using prox switches, and then sending it to a pre-determined part of the table and cutting down, only to find it landing randomly in a circle about .1" in diameter each time (not distinctly spread out in X or Y). That's almost ten times more slop than would be acceptable for the work it needs to do to be widely used; we just don't handle too many jobs that are both large-format and that sloppy. Is that a normal amount of error? Are there any good tricks for cutting that CEP down to the order of a few mils at most?