I have a j-head b-port that I am retrofitting the original anilam controls with emc2 and mesa 5i20 and 7i33 boards. The machine uses the original anilam/ accurite linear encoders for feedback. I've got the machine running very well at this point and have gotten past the regular servo tuning issues. My issue is with actual machine movement versus commanded machine movement.
I will focus on the z-axis for now, because the encoder resolution is still readable on the linear encoder. The label reads .0005". Which would make the input scale 2000. These settings cause a grossly innacurate movement on the z-axis. If I command a jog of 1", the axis only moves .885" or so, but the dro readout reads in .0005 increments. If I change the input scale to 2539, the axis moves 1" when comanded. But when i command a 2" movement it stops short by about 3 or 4 thou. Also the dro in axis reads in .0004 increments (which really bothers me for some reason.)
Cliff notes:
An input scale corresponding to actual glass scale resolution of .0005" (2000) produces a move that is short by .200" or so. But results in dro display of
.0005 increments (which is the way i like it)
An input scale of 2539 gets me really close to a 1" movement, but seems to be off at 2" movement. But also gives me .0004" increments in axis dro.
What am i missing here? has everyone else had to modify what they think is the right input scale to adjust for the actual posistion of the movement? I guess I can live with increments of .0004" instead of .0005", but it just looks and seems wrong to me.
clearly there is something wrong with your numbers (inputs). 2539 sounds suspiciously close to some multiple of 25.4. It has been a really, really long time since anyone made glass scales in anything other than metric increments. I suspect that is where your problem is. You can set it up to be correct, you just have to figure out what is wrong.
Ok that makes sense. Doing the math for a .01mm resolution it comes out to be 2540. I guess I'll just have to live with the .0004 increments, unless I can find some other way around it. Thanks
Incorrect actual joint movement
Reply with Quote
