On 4/10/2013 10:01 AM, Daniel Rogge wrote: > I'd like to weigh in with the following test: > > Running LCNC 2.5, copy the sim/axis config to your local configs, then change > the max_acceleration for axis 0, 1, and 2 to 1.0 (previously 100): > > MAX_ACCELERATION = 1.0 > > Then run both of the following programs: > > % > (1 inch square) > G90 G54 G20 > G64 P.125 > G0 X0 Y0 Z0 > G1 Y1 F50 > X1 > Y0 > X0 > m30 > > > > % > (1" square with rounded corners) > G90 G54 G20 > G64 > G0 X0 Y.25 Z0 > G1 Y.75 F50 > G2 X.25 Y1 I.25 > G1 X.75 > G2 X1 Y.75 J-.25 > G1 Y.25 > G2 X.75 Y0 I-.25 > G1 X.25 > G2 X0 Y.25 J.25 > M30 > > The tool path on both programs is nearly identical because the square with > hard corners is run at G64 P.125. > > http://static.inky.ws/image/3839/Screenshot-rounded_square.ngc%20-%20AXIS%202.5.0%20on%20LinuxCNC-HAL-SIM-AXIS.png > > If you watch the velocity display you will see that the program with only > line-line transitions (1" square) reaches 35 inches/min - while the program > with line-arc or arc-line transitions (1" rounded square) runs at only 26 > inches/min. > > The arcs are certainly tangent here, and no one can point blame at the CAM > software (although you're welcome to point blame at my poor hand coding > style). > >
Daniel: I really like that you have bounded the argument by introducing these two simple test files and using them to compare LinuxCNC behavior to LinuxCNC behavior. On a computer close at hand at my desk I happen to have available a virtual host running Ubuntu 10.04LTS and LinuxCNC2.5.2-189...(a relic of some previous testing) so I ran your two test files. I get the same results you do. Just for fun, on this same computer I installed yet another virtual host running Ubuntu 8.04LTS and EMC2 2.3.0 (installing from the old ubuntu-8.04-desktop-emc2-aj13-i386.iso <http://dsplabs.upt.ro/%7Ejuve/emc/get.php?file=ubuntu-8.04-desktop-emc2-aj13-i386.iso>). *Again* I get the same results for your two test files: 35+ ipm for the square and 26+ ipm for the rounded square. Evidently, this behavior precedes 2.4. All: I am a total ignoramus when it comes to the trajectory planning and motion control aspects of LinuxCNC. From my seat in the peanut gallery, it seems there is a divide between those who believe we have a proper set of algorithms properly implemented that have been tested successfully and those who believe this reported behavior must mean either the implementation is deficient or the algorithms imperfect (or both!). So far, the two groups of discussants seem to have been talking past each other, both in the 2011 exchanges and now. The following questions are simply my noodling and not any attempt to state a personal position: Do Daniel's two files constitute a valid test? That is to say, should one expect substantially the same behavior when each is executed? If the answer is yes, then why isn't it the behavior the same? If the answer is no, then why should it not be the same behavior? Once the dust has settled I hope the subject can be explained more fully in the LinuxCNC documentation. For some time we have advertised "constant velocity control" as a feature of EMC/LinuxCNC (usually in an "About..." section) but try searching on "constant velocity". Two technically meaningful pages on the Wiki respond to this search term: http://wiki.linuxcnc.org/emcinfo.pl%3FTrapezoidal_Velocity_Profile_Trajectory_Planner and http://wiki.linuxcnc.org/emcinfo.pl%3FSimple_Tp_Notes. Pardon me, but if this is such an important topic why is it buried? The rest of the docs don't waste many words, as my grandmother used to say, about feed or speed. Just my 2cents worth. Regards, Kent ------------------------------------------------------------------------------ Precog is a next-generation analytics platform capable of advanced analytics on semi-structured data. The platform includes APIs for building apps and a phenomenal toolset for data science. Developers can use our toolset for easy data analysis & visualization. Get a free account! http://www2.precog.com/precogplatform/slashdotnewsletter _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
