I have also found problems with curve->line and line->curve transitions. My machine is used to paint lines and uses a fourth tangential axis to keep the brush tangent to the tool path. I get significant decelerations/accelerations on these transitions and it shows up as aberrations in the paint stroke. The G code is generated using biarc approximations and all line/curve transitions are tangential (at least to a fairly high degree of precision.) I've tried using different G64 values but it doesn't help much and the tool path precision gets way off. I was going to order a 5i25 card to see if maybe improving the stepgen performance would help, but I'm no CNC expert. Anyway, it's good to know I'm not alone in seeing this behavior. Thanks, - Claude
On Wed, Apr 10, 2013 at 4:00 PM, Gregg Eshelman <g_ala...@yahoo.com> wrote: > On the water jet site (I want to think it was CMC, but memory is hazy) the > comparison between their full path pre scanning and short distance > lookahead was that the full pre scan could adjust to things like a long > straight followed by a series of curves and short straights by slowing down > to a best possible average speed through the twisty section, still managing > to speed up a bit where possible. > > Short distance lookahead is always getting surprised by the next curve, > causing rapid attempts to change speed and resulting in lower cut quality > and more wear and tear on the machine. > > Compare it to a skilled race car driver who has made a practice lap and > committed every twist and turn to memory, planning out all the gear changes > and how fast to go at all points along the course. > > Short range lookahead is like a semi-skilled driver who has never been on > the track before and starts the race without even a practice lap. He can > see a turn or two ahead and has the skill to not run off the track, but > will try to go as fast as possible all the time and is always having to > stomp on the brake at every turn. > > The skilled driver maintains a smooth speed profile through a chicane > while the amateur comes screaming in, hits the brakes hard and has to putt > through the curves slowly to avoid spinning off the track. > > The driver who runs the course before the race ends up lapping the track > faster, with less wear on the consumables. > > So why not have a "practice lap" simulation that can generate a complete > accel/speed/decel profile for a toolpath then feed that data to LinuxCNC? > > > ------------------------------------------------------------------------------ > 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 > ------------------------------------------------------------------------------ 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