using HAL to correct axis errors

Mike_Eitel
Offline
Platinum Member

18 Feb 2018 08:47 #106115 by Mike_Eitel

Replied by Mike_Eitel on topic using HAL to correct axis errors

I like your true diy

Please Log in or Create an account to join the conversation.

pgf
Offline
Premium Member

18 Feb 2018 15:14 #106135 by pgf

Replied by pgf on topic using HAL to correct axis errors

I think "true diy" might be a euphemism for "doesn't work all that well". :-)

Please Log in or Create an account to join the conversation.

Mike_Eitel
Offline
Platinum Member

18 Feb 2018 15:24 #106136 by Mike_Eitel

Replied by Mike_Eitel on topic using HAL to correct axis errors

But its fun to think about and tinker.
Have fun!

Please Log in or Create an account to join the conversation.

andypugh
Offline
Moderator

28 Feb 2018 17:22 #106725 by andypugh

Replied by andypugh on topic using HAL to correct axis errors

There is also a special kinematics file that can be used to correct a skewed machine:
wiki.linuxcnc.org/cgi-bin/wiki.pl?Contri...y_XY_skew_correction

That might not work so well with a gantry machine, it predates joints-axes.

Please Log in or Create an account to join the conversation.

pgf
Offline
Premium Member

28 Feb 2018 17:32 #106728 by pgf

Replied by pgf on topic using HAL to correct axis errors

thanks andy. guess that could have saved me a bit of work. oh well -- it wasn't hard, and i enjoyed it. and what's most important....

MY MILL CAN CUT RIGHT ANGLES!!!

there. just had to say that. :-)

paul

Please Log in or Create an account to join the conversation.

Mike_Eitel
Offline
Platinum Member

28 Feb 2018 19:03 #106739 by Mike_Eitel

Replied by Mike_Eitel on topic using HAL to correct axis errors

Nice to read from a happy man!

The following user(s) said Thank You: pgf

Please Log in or Create an account to join the conversation.

LAUSCH
Offline
Junior Member

04 Sep 2022 09:43 #251158 by LAUSCH

Replied by LAUSCH on topic using HAL to correct axis errors

Hi paul,

Can you explain why you set:
option singleton yes;
If you do this you will get errors from the lines in HAL:
addf nonsquare-table.0.cmd servo-thread
addf nonsquare-table.0.fb servo-thread

I have thus comment the line
//option singleton yes;
And LinuxCNC starts without error.

Cheers
Christian

Please Log in or Create an account to join the conversation.

pgf
Offline
Premium Member

04 Sep 2022 18:00 #251178 by pgf

Replied by pgf on topic using HAL to correct axis errors

Well, I clearly didn't get errors from that code when I last ran the mill. I had to look up what the "singleton" option does -- it seems appropriate for this module, since you can really only have one of these components per system. But perhaps you're running a newer version of hal, with tighter restrictions on naming.

paul

Please Log in or Create an account to join the conversation.

pgf
Offline
Premium Member

05 Apr 2025 14:57 #325765 by pgf

Replied by pgf on topic using HAL to correct axis errors

I wanted to bring this thread up to date, in case anyone finds it and finds it useful. I recently upgraded my mill from 2.7.4 to 2.9.4, which of course brought in the joints axes changes.

Those required some rework of my skew corrector. The component module itself was unchanged. It's in a post above .

But the linkage into HAL did change, due to joints axes, and the introduction of pid.

First, 2.7: the original HAL had a fairly simple connection between motor and stepgen:

  ypos-cmd
      axis.1.motor-pos-cmd ==> stepgen.1.position-cmd
  ypos-fb
      stepgen.1.position-fb ==> axis.1.motor-pos-fb

  xpos-cmd
      axis.0.motor-pos-cmd ==> stepgen.0.position-cmd
  xpos-fb
      stepgen.0.position-fb ==> axis.0.motor-pos-fb

I was able to simply insert the nonsquare function block into the middle of each Y axis signal path, and tap into the X signal path.

 ypos-cmd
      axis.1.motor-pos-cmd ==> nonsquare-table.y-in
  ypos-cmd-adjusted
      nonsquare-table.y-out ==> stepgen.1.position-cmd
  ypos-fb-adjusted
      stepgen.1.position-fb ==> nonsquare-table.y-fb-in
  ypos-fb
      nonsquare-table.y-fb-out ==> axis.1.motor-pos-fb

  xpos-cmd
      axis.0.motor-pos-cmd ==> stepgen.0.position-cmd
                           ==> nonsquare-table.x-in
  xpos-fb
      stepgen.0.position-fb ==> axis.0.motor-pos-fb
                            ==> nonsquare-table.x-fb-in

The newer LinuxCNC path in 2.8/2.9 is a little more complicated, with the introduction of the PID mechanism.

  y-pos-cmd
      joint.1.motor-pos-cmd ==> pid.y.command
  y-output
      pid.y.output ==> hm2_7i96.0.stepgen.01.velocity-cmd
  y-pos-fb
      hm2_7i96.0.stepgen.01.position-fb ==> joint.1.motor-pos-fb
                                        ==> pid.y.feedback

  x-pos-cmd
      joint.0.motor-pos-cmd ==> pid.x.command
  x-output
      pid.x.output ==> hm2_7i96.0.stepgen.00.velocity-cmd
  x-pos-fb
      hm2_7i96.0.stepgen.00.position-fb ==> joint.0.motor-pos-fb
                                        ==> pid.x.feedback

Integrating the nonsquare function in the new HAL is similar, butasymmetric. TreIn the output direction, the nonsquare function is inserted between the joint and pid (which then feeds stepgen), but in the feedback direction, we insert between stepgen and the joint.

Regardless, the new function is inserted just after (cmd) and before (fb) the joint command. The X control for the new function is similarly attached directly to the joint function.

  y-pos-cmd
      joint.1.motor-pos-cmd ==> nonsquare-table.y-in
  new-y-pos-cmd
      nonsquare-table.y-out ==> pid.y.command
  y-output
      pid.y.output ==> hm2_7i96.0.stepgen.01.velocity-cmd
  y-pos-fb
      hm2_7i96.0.stepgen.01.position-fb ==> nonsquare-table.y-fb-in
                                        ==> pid.y.feedback
  new-y-pos-cmd-fb
      nonsquare-table.y-fb-out ==> joint.1.motor-pos-fb

  x-pos-cmd
      joint.0.motor-pos-cmd ==> pid.x.command
                            ==> nonsquare-table.x-in
  x-output
      pid.x.output ==> hm2_7i96.0.stepgen.00.velocity-cmd
  x-pos-fb
      hm2_7i96.0.stepgen.00.position-fb ==> joint.0.motor-pos-fb
                                        ==> pid.x.feedback
                                        ==> nonsquare-table.x-fb-in

The following user(s) said Thank You: tommylight

Please Log in or Create an account to join the conversation.

Time to create page: 0.079 seconds