General questions on best method for installing Linuxcnc for software stepping.

Yeah I get you there I am a beginner when it comes to Linux as well basically there was a script that I used ( I think I found it in a forum post) in order to set the RTAI kernel boot paramiters because if you did it manually you would need to change 2 separate files. In a few hours when I get back to the machine I'll post the script.

MDM3D wrote: I think that my next step is to learn about gantrykins.c I tried using millkins.c to account for skew but couldn't get it to load with the latest dev version, I got it to compile though.

If you are using the latest dev version then gantrykins doesn't exist any more. The same functionality is included in trivkins.

There is an example of what a gantry config with trivkins should look like here, look particularly in the [KINS] section of the INI
github.com/LinuxCNC/linuxcnc/blob/master.../sim/axis/gantry.ini

Thanks Andy
So I read the gantry.ini and I see how the axis and joints are defined but dumb question how would I go about setting the skew paramiters some setp style syntax paramiters or am I missing something? Also is there a way to map each axis to account for nonlinear errors. I have read a good chunk of the documentation manual but I probably missed how to store probe data for long term use.

There isn't any skew correction in trivkins, and illkins doesn't do gantries.

But what is the source of the skew? One of the main points of the new gantry handling is that with home-switches on each side of the gantry it will auto-square during homing.

Can you give us a lot more detail about what problems you have that need to be adressed?

To clarify, the gantry is squared during the homing sequence. In rough terms, the gantry travels towards the home switches and once it hits the first one, that side stops moving and the other keeps travelling until it hits its own switch at which pont the gantry should be square.

Of course its a bit more complicated than this as there are a couple of steps in the homing sequence but the description above sums it up.

Basically when I built my machine I did the best I could to build it as square and orthogonal as I could but I have a small squareness problem I'm off at about .oo4 over 24 inches of travel but I also have a .003 waveyness to my x axis(the gantry) unfortunately everytime I adjust the linear skew to 0 my tram goes out the window and because I am using double ball nut ballscrews my backlash is approximately .0001 I figured I could store a correction into each axis to account for both errors possibly using a table of probe data

Ah I see the problem I only use one motor per axis I put the ball screw in the middle of the table and then ran a bar from each side of the axis to tie the carriages together so the skew is completely a build error and not a motor synchronizing issue. The picture in the first post shows my setup.

#make a new text file with this in it.
sudo cp 07_rtai /etc/grub.d
sudo chmod +x /etc/grub.d/07_rtai
sudo update-grub
#
then modify etc/grub.d/07_rtai using the options listed in 07_rtai
finally run the script as a program and reboot
run the latency test and tweak until satisfied


my particular settings look like this

#MAV: uncomment isolcpus=1 for dual core, isolcpus=1,2,3 for quad
# these both assume hyperthreading is already disabled in bios
#
# optionally uncomment nohlt and idle=poll lines to experiment with those
# note: try idle=poll first as nohlt may cause increased power usage
#
#RTAI_OPTS="isolcpus=1"
#RTAI_OPTS="isolcpus=1,2,3"
#RTAI_OPTS="nohlt ${RTAI_OPTS}"
RTAI_OPTS="idle=poll ${RTAI_OPTS}" # this is the polling option that worked for me.
GRUB_CMDLINE_LINUX_RTAI="${GRUB_CMDLINE_LINUX_DEFAULT} ${RTAI_OPTS}"

ignore this post

MDM3D wrote: Ah I see the problem I only use one motor per axis I put the ball screw in the middle of the table and then ran a bar from each side of the axis to tie the carriages together so the skew is completely a build error and not a motor synchronizing issue.

OK, in that case you should be able to fix the problem with millkins, as I think you already said?

wiki.linuxcnc.org/cgi-bin/wiki.pl?Contri...y_XY_skew_correction

Have you tried it? What was the result?