#linuxcnc-devel Logs

Jul 15 2017

09:37 AM jepler: Even if interpolationRate works as designed, I doubt it's useful. I'd love to see some hard info to the contrary.
09:39 AM jepler: Basically: say you want to run a 1ms servo cycle, but your kinematics takes 1.5ms. You can't do it, because you miss your realtime deadlines. "aha" you say "just set things up so that interpolationRate is 10"
09:39 AM jepler: but that doesn't fix anything, it just gives you 1 1.5ms servo cycle followed by 9 short ones
09:39 AM jepler: to actually work, TP and kinematics need to be in one thread, and interpolation, pid, etc., have to be in a faster thread
02:23 PM seb_kuzminsky: jepler: i agree, some numbers would be good here
04:18 PM skunkworks: jogged the k&t into the tomb stone.. pushed it .050"
04:19 PM skunkworks: removed the .1 jog postion on the jog selector....
09:27 PM skunkworks: http://www.machsupport.com/forum/index.php/topic,35157.0.html
09:43 PM pcw_home: I guess real time is not really needed for the kinematics/reverse kinematics as long there's no need to deviate from the pre-planned trajectory
09:43 PM skunkworks: doing it all in gcode seems like a challenge..
09:44 PM pcw_home: Yeah thats ugly
09:46 PM pcw_home: just saying that I think a Mach like buffered system could implement this without real a time TP
09:53 PM pcw_home: I think Mach must have lost a huge amount of market share with the rough transition to Mach4
09:58 PM skunkworks: that is my feeling... people are hanging on to mach3
10:05 PM skunkworks: or going to something else like planet cnc or uccnc
10:17 PM pcw_home: Yeah I think ucccnc is the main one
10:18 PM pcw_home: (for low end anyway)
10:27 PM skunkworks: right - direct replacement of the printer port pretty much with a usb interface
10:27 PM skunkworks: (I think they have some ethernet solutions too)