You must have a substantial amount of backlash? The backlash is taken out at max axis acceleration. that is why you need to have 2X acc for the stepgen_maxaccel. There is a chance that the axis could be accelerating and the backlash taken out at the same time causing 2x the acceleration being sent to the stepgens.
On Wed, Aug 28, 2019 at 9:24 PM John Dammeyer <jo...@autoartisans.com> wrote: > Gene, > > You've got it backwards. The parallel port version is correct which is > why that line is there. > > > setp parport.0.pin-07-out-invert 1 > > remove the above line, or set it false > > > net zdir => parport.0.pin-07-out > > Peter Wallace suggested I do this: > "If you are using LinuxCNC 2.8 you can set the stepgens dir polarity > parameter." > > Can't find a "dir polarity" parameter here: > http://linuxcnc.org/docs/html/drivers/hostmot2.html#_parameters_3 > > However there is an "invert_output" parameter . Still nothing I can find > online really guides me in the direction of doing it with port inversion or > how to use that invert_output since the dmesg instructions in the above web > link don't help. > > However the other suggestion from Peter which was to set the stepgen scale > to minus. Did that to both Z and Y axis and now homing works. > > I suspect because the parallel port implementation always reported > "unexpected real time delay" on start-up that it wasn't able to actually > keep up with the acceleration parameters in the .ini file. Now with the > 7i92H, the acceleration on the Z axis is way too high for the Gecko run > stepper motor. It's not losing steps but doesn't sound happy. A bit of > tweaking with values and it's much smoother. > > I've also played around with the Y axis acceleration since it too has a > lot of mass to move; not only the X axis but also the X axis table > rotation mass. That's why the DC Servo has 4:1 instead of 3:1 reduction > like the X. Max ipm is therefore 120ipm for Y and 150ipm for X. For the > knee with the Gecko and 3:1 to the bevel gear I get 24ipm. Any higher than > that and the stepper will skip. > > The backlash compensation is also far harsher. A series of jogs in the > same direction create almost no vibration on the head of the mill (hand > placed against it) but change direction and my hand feels the bump as the > backlash is first taken up. This was especially noticeable on the knee > with the slight bevel gear connection to the knee screw until I dropped > acceleration to a much lower value than I had on the parallel port version. > > The 7i92H requires the STEPGEN_MAXACCEL to be double the > MAX_ACCELERATION. Setting it lower than that caused strange motion > problems so I suspect the only way to get rid of the hard bump as backlash > is taken out is to also reduce axis acceleration. > > Or buy ball screws which is on the list... > > John > > > > > > > > > > > _______________________________________________ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
