> > > > Thanks Peter, > > I've got pulses coming out the Spindle Pins on connector P1. > > While snooping with the scope and looking at the generated signals I've > run into an issue. I haven't looked at any other axis but the following > information has to do with the X axis. The attached screen shot shows the > pulses as I let off on the jog button for the X axis. > > > > Scope trigger is on the direction signal (blue trace) falling edge. > > Notice after the 2uS step pulses (yellow trace) have stopped the direction > signal changes 3 times. > > Why? > > This has to do with the stepgen control loop having much greater > resolution than a single step so tiny corrections may change the > direction (though timing is always obeyed). It may also have to do with PID > setup. >
It doesn't appear to do any harm since the step lines are stable. But it's odd and very repeatable in either direction. The direction signal always returns to + direction regardless of the jog direction. The attached screen shot show a 0.001" move in the negative direction with exactly 15 steps which matches motor encoder and 3:1 pulley and 5 TPI lead screw. I'll temporarily remove the limit switches, backlash and run a short G-Code program that moves 0.001 and then -0.001" and capture the direction signal. Very curious. > > > > > > I'm also seeing occasional joint 0 following errors when I change direction > which is what prompted putting the scope on the signals. It's open loop > step/dir with no encoder feedback so something is not quite right. > > > > try commenting out these lines (for all axis) > > net x-vel-cmd => pid.x.command-deriv > setp pid.x.maxerror .0005 > > This didn't make any difference. What does fix it is disabling BACKLASH. And that led me to the STEPGEN_MAXACCEL set too low. Initial testing didn't have backlash enabled so the 25% increase in STEPGEN_MAXACCEL was more than adequate. But with backlash enabled it broke. Should really follow the comments in there. Turns out on the actual machine with the parallel port I did have STEPGEN_MAXACCEL set to just over 100% larger. > originally created by pncconfig. Should not > > setp hm2_7i92.0.stepgen.00.control-type 1 > > be control-type 0? > > > > No, needs to be 1 So what does it need to be for the spindle then? The tutorial suggests stepgen is type 0 for step/dir and "ctrl_type" type v for velocity. I haven't found the details on "control-type" in a document yet. hm2_7i92.0.stepgen.05.control-type Perhaps I overlooked it somewhere? loadrt stepgen step_type=0,0 ctrl_type=v,v
_______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users