#linuxcnc-devel Logs

Jul 28 2018

10:44 AM seb_kuzminsky: http://linuxcnc.org/docs/devel/html/gcode/g-code.html#gcode:g10-l1
10:46 AM seb_kuzminsky: oops, i meant http://linuxcnc.org/docs/devel/html/gcode/g-code.html#gcode:g10-l2
10:46 AM seb_kuzminsky: down in "Important Concepts" it says: "When a rotation is in effect jogging an axis will only move that axis in a positive or negative direction and not along the rotated axis.
10:46 AM seb_kuzminsky: is that true? don't most guis do axis jogging on all kins nowdays?
10:46 AM hazzy-lab: seb_kuzminsky: In my experience that is true
10:50 AM seb_kuzminsky: huh, yep, it axis-jogs in the un-rotated coordinate system
10:51 AM hazzy-lab: Right, it would be cool if the jog axes rotated too, but could also be a little disorienting, lol
10:51 AM seb_kuzminsky: we obviously need a third kind of jogging: joint, axis, and rotated-axis
10:52 AM hazzy-lab: That would be sweet
07:44 PM andypugh: So, locking-indexer axes. Where did the current behaviour originate?
08:14 PM seb_kuzminsky: andypugh: i think cradek wrote it some years ago
08:14 PM andypugh: Yes, Gitgub is of the same opinion.
08:15 PM andypugh: I am trying to work out if it is currently broken or if it never worked for linear axes (and wasn’t meant to)
08:15 PM seb_kuzminsky: i think it was written for locking rotary joints, and has never known anything about linear joints
08:16 PM andypugh: It’s pretty similar to take the sim-axis config and add a LOCKING_INDEXER_JOINT for each axis and LOCKING_INDEXER for each joint. Then you can watch halmeters and set is-unlocked pins and get pretty much expected behaviour.
08:17 PM andypugh: But once homed, eberything seems exactly like normal.
08:17 PM andypugh: (pretty simple)
08:18 PM seb_kuzminsky: so before homing all joints get "locking" behavior, but after homing only rotary joints get locking behavior?
08:18 PM seb_kuzminsky: sounds like a "free" vs "teleop/auto" thing in Motion
08:18 PM andypugh: My sim had no rotary axes, so I can’t answer the second part.
08:19 PM seb_kuzminsky: if you home, then switch to joint mode, do you get locking behavior back on all joints?
08:20 PM andypugh: How do I switch to joint mode?
08:21 PM andypugh: (I know it should be $ but that seems to not be working)
08:21 PM seb_kuzminsky: i also think it's $ in axis
08:21 PM seb_kuzminsky: if that's not working then i dont know :-
08:21 PM seb_kuzminsky: :-/
08:22 PM andypugh: Maybe no longer an option in 2.8 and trivkins.
08:22 PM seb_kuzminsky: if we broke joint mode, that's a bug
08:23 PM andypugh: Pity we can’t remap G0 and G1 to wrap them in engage-disengage logic
08:23 PM andypugh: If you have LinuxCNC there, try sim/axis_mm. I don’t seem to be able to get joint mode.
08:24 PM andypugh: It could be a keyboard thing, I am using a UK Mac keyboard and LinuxCNC in a VM.
08:27 PM seb_kuzminsky: i don't get joint/world mode in axis on trivkins machines
08:28 PM andypugh: That is probably not wrong.
08:29 PM seb_kuzminsky: but i *do* get joint/world mode on non-identity kins (gantry) in master
08:29 PM seb_kuzminsky: so i guess... that's ok
08:30 PM andypugh: Locking indexer does seem very close to being right for a shared-motor machine.
08:31 PM andypugh: There is a picture of the machine here:
08:31 PM andypugh: https://forum.linuxcnc.org/12-milling/34239-servo-recomendation-whn-11?start=10#114870
08:32 PM andypugh: (attached, not embedded)
08:32 PM andypugh: Keeping the one big shared servo doesn’t actually seem crazy in that case.
08:33 PM andypugh: (Slightly sad photo, looks like a big business being stripped out_
08:39 PM dgarr: andypugh: for trivkins,use kinstype=both to allow joint/world switching (man trivkins)
08:42 PM dgarr: first commit ( i think) for locking indexers is clearly rotary (abc) only: https://github.com/LinuxCNC/linuxcnc/commit/5a41a2bf0
08:44 PM andypugh: He went..,
08:44 PM andypugh: Joint mode is no help here anyway
09:32 PM seb_kuzminsky: thanks andypugh!
09:36 PM andypugh: I need to fix carousel. Probably tomorrow. I broke it at the weekend. It now supports more changer types, but doesn’t support the machine I was using at the weekend any more. Ironically the machine it now works with, and the machine it no longer works with, were less than 1000mm apart in Stuttgart.
10:53 PM hazzy-lab: seb_kuzminsky: Thanks for the G38 doc clarifications!
10:56 PM hazzy-lab: #5081-#5089 for ABS probed pos, yippee!
10:57 PM hazzy-lab: The LCNC team is awesome!
10:58 PM Tom_L: is that new?
11:03 PM hazzy-lab: That the team is awesome? No :)
11:03 PM hazzy-lab: Yes, seb just started working on it
11:05 PM hazzy-lab: I complained that the probed pos parameters were in the current WCS instead of machine coords (as in most controls), so he is adding #5081-#5089 for the ABS pos at which the probe tripped