What happens if you plot motion.current-vel and Chris's Fcode pin in halscope? Really should not be using the eyeometer here....
And yes, the feed values are tokenised and thrown away before it gets to motion. Rod Webster *1300 896 832* +61 435 765 611 VMN® www.vmn.com.au On Wed, 1 Apr 2020 at 14:47, Reinhard <reinha...@schwarzrot-design.de> wrote: > Hi Rob, > > On Mittwoch, 1. April 2020, 05:42:08 CEST Rod Webster wrote: > > I'm a bit confused by Reinhard's views. > > We spent a fair bit of time working though this with plasma machines > (where > > knowing the commanded velocity is critical) and I don't recall seeing an > > instance where commanded velocity was exceeded. BUt that was a while ago > > Well, the testcase came from Chris and it is the file 3D_chips.ngc > modified so > that some lines contain different f-words. > That file takes long motion time, so discrepancy between readahead and > real > motion comes to light. > > I added a F-word with a slow value (like F100) to the beginning and a > faster > one (like F400) to the end. Without changing the feed overwrite, the > machine > starts to run with F400 as feed limit - which is aparently wrong. > > I run linuxcnc with axis (to have visual feedback about motions) and my > app as > a status browser, which shows the actual feed rate from status feedback > channel. Status feedback reflects accelerating and decelerating ... > ... so its pretty obvious, that the feed values come from motion and not > from > interpreter. > > > cheers Reinhard > > > > > _______________________________________________ > Emc-developers mailing list > Emc-developers@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-developers > _______________________________________________ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers