Backlash move midway on linear move since update

More
30 May 2016 15:40 #75248 by NHGeorge
Hello new to forum but have been using emc2 for years and just upgraded from 2.7.3 t0 2.7.4 and was contouring a simple rectangle this morning and every time i was moving in y axis on both ends the machine makes a backlash move mid path and leaving a pretty good step. used g61 prior to this upgrade and noticed g61.1 is now exact stop code? tried both and g64 p.001 and slowed feed etc but still get backlash type move. never got this prior to upgrade.
thanks G

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

More
02 Jun 2016 02:31 #75317 by Todd Zuercher
Could you post your problem g-code as an attachment?

Is your machine setup to use backlash compensation?

Can you measure the offset the move made?

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

More
03 Jun 2016 13:49 #75426 by NHGeorge
Hi Todd
Thanks I don't think i saved the code but ill take a look later but measured the step and approx .013 and i believe thats also what i have for my backlash set at. The code was pretty simple just my safety line and repeated contour moves at .100 depths that i have used for numerous jobs prior. I did try moving the g61 down and tried g61.1 and g64 p.001 as well as changing feed and was still getting this glitch mid path and remember at some point depending on feed i would get on one end and not the other and at some point in my testing the y would do the same. I feel there is an error in the emc coding dealing with backlash and trajectory planning. I did find in my search for an answer that trajectory planning for this release was a new design vs tweaks to the previous version.
Thanks G
%
g0 g90 g40 g80 g61
g0 g54 x-.25 y.125 z.1
g01 z-.1 f10.
x6.025 f20.
y-1.825
x-.125
y.125
z-.2 f10.
x6.025 f20.

etc..

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

More
04 Jun 2016 15:19 #75490 by andypugh
The changes from 2.7.3 to 2.7.4 are listed here:
www.linuxcnc.org/2016/02/08/release-2.7.4/

There were no changes to motion at all.

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

More
05 Jun 2016 12:06 #75519 by NHGeorge
Hi Andy for some reason I thought I read the traj planner changed. I did some testing and had remembered i had installed Qjoypad to use my game controller as mpg. I unplugged and disabled and test still showing moves mid path, the only thing i can find is if i disable backlash in my ini this stops happening. I attached my files as maybe someone can see something wrong in them.
Thanks G
Attachments:

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

More
07 Jun 2016 10:32 #75618 by andypugh
is it possible that your actual backlash has increased or decreased?

It could just be the part. LinuxCNC only knows that it has changed direction. It has no way to know that the cutting force has changed direction. So for some parts you will see a move from one side of backash to the other that LinuxCNC has no way to compensate for.

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

More
07 Jun 2016 21:15 #75661 by NHGeorge
I know what your saying and I can watch the machine/steppers make the move while not touching the part sometimes based on speed and length of the passes it may happen on all four sides or just one and is repeatable. I have redone my step config and only thing at this point that is changed is the latency figure is 3k less than before 16k-13k. I am going to test and work my way with the new ini file back into the backlash and following error changes that previously worked and try to find more clues.

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

More
08 Jun 2016 07:35 #75673 by Rick G
As you are making air cuts for testing perhaps dramatically increase the backlash setting to make it easier to see what is going on and see it the mid way move also increases,

Rick G

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

More
08 Jun 2016 11:44 #75682 by LearningLinuxCNC
Is it possible that the G54 coordinate system is rotated by some very small amount causing a very small y move when a straight x move is requested therefore kicking in the backlash compensation. Same would occur on a long enough move in the y direction with a small step in the x direction.

You might try resetting the G54 coordinate system so that it has 0 rotation. I am just grasping at straws here but it appears that it could cause the behavior you are talking about.

Jim

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

Time to create page: 0.075 seconds
Powered by Kunena Forum