Jon Elson wrote: > Brian wrote: > >> This is what I am currently doing, but it doesn't really do much for >> me. I even have a bar graph of the current position error. I think >> it would be a natural progression for EMC to have the option to use >> feedback from a source that is subject to screw comp, or one that >> isn't. >> >> Brian >> >> On Fri, Aug 6, 2010 at 1:51 PM, Andy Pugh<[email protected]> wrote: >> >> >>> It is in the right place if the position feedback component is the >>> motor encoders. >>> >>> > OK, I still don't understand the fundamental problem. My understanding > was that the screw error was applied to the position feedback, so the > position displayed on the screen is already corrected for error. I > think Andy is confirming this above. > > So, if that is true, then you are using the wrong motion component, > because it is one that can't handle actual position feedback. If you > want to disable the motors, move the machine and then re-enable, I think > you MUST change the way you command motion, to use one of the possible > schemes that look more like a servo system. EMC1 did this, with a > software simulation of a step-servo. EMC2's stepgen is decidedly in the > open-loop mode, and so is inappropriate for this type of use. I'm sorry > that it seems like I'm plugging my hardware, but I know that I have had > this capability since 2002 with my Universal Stepper Controller. I'm > pretty sure the Mesa boards can do the same thing. I think this > functionality has been taken out of EMC2 because it confused people. > Since all the parts are there, it probably would not be that hard to put > back in, there may even have been an early EMC2 step generator component > that followed the old servo mode that could be resurrected. > Nothing has been removed, you would just use the stepgen in velocity mode, with a PID driving it (making FF1=1 should give you the equivalent of open-loop mode, plus or minus a step here and there).
You can then add in some I, so that the stepgen eventually ends up where the scale says it's in the right place. The real problem is that screw mapping is meant for systems that have the screw between the feedback device and the table, but this machine isn't like that since the scale is actually on the table. You can't drive the PID very well with only scale feedback since it's too loosely coupled to the motors, but that's irrelevant since you can't really use steppers like servos anyway. I'd suggest changing the stepgen setup so you have PID driving a velocity mode stepgen, with FF1=1 and a little bit of I also. Get rid of the screw mapping. Use the scale feedback for motor-pos-fb (and the feedback input to the PID). That will track manual motion while in the machine off state, and won't jump anywhere when the machine is turned back on. (at least it shouldn't anyway - YMMV) - Steve ------------------------------------------------------------------------------ This SF.net email is sponsored by Make an app they can't live without Enter the BlackBerry Developer Challenge http://p.sf.net/sfu/RIM-dev2dev _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
