unpredictable feed speeds

My machine is a xyuv foam cutter on raspi 4, using linuxcnc iso, and a mesa board.

I have a file generated by wing g-code.
The gcode, ini and hal are attached.

The problem is that a couple of the moves go much faster than the rest, even though it's just a bunch of G01's without F's. Too fast for the hot wire.



The first is right at the beginning
G21
G90
G49 G64 P0.01
(seek to start height)
G00 X-5.000 Y20.244 U-5.000 V20.052 F200.0
G01 X0.000 Y20.244 U0.500 V20.052   ;; this line changes behavior
G01 X0.000 Y20.244 U167.500 V20.052

The code goes quickly to a start position. It's G00, that's what it's supposed to do.
With the next line, the one with the comment, things proceed at a reasonable speed. In particular the one after, going to U167

If I remove the line with the comment, the move to U167 goes way faster, though still slower than rapids.

The work around of adding the line would be ok except it happens again at the end of the program, when it races from U267 back home.

G01 X0.000 Y17.052 U267.500 V17.244
(close trailing edge)
G01 X0.000 Y20.244 U267.500 V20.052
(retract out of foam)
G01 X-5.000 Y20.244 U-5.000 V20.052
G4 P0.075
G00 X-10.000 Y10.000 U-10.000 V10.000


But why are the feed speeds changing at all?

Please help
Miles



 

Search for inverse time or something like that, for a 4'th axis, it explains also the U and V axis.

I wrote a script to fix this. The problem is that the feed rate considers only the XY plane not the UV plane.
So the script is set up as a filter and converts the G0cde to G93 feed mode.

This thread explains it all, and contains the script.
www.forum.linuxcnc.org/38-general-linuxc...tcher-for-me?start=0

Make sure you find the final version....

That was it. Thank you.

In the thread linked by Andy, I found two versions. The original by Andy and an iteration on it by Alek. Both had some issues (mostly minor) that I've fixed and will post in the other thread.

Thank you for taking the time to help get me going.

Miles