#linuxcnc-devel Logs

Jun 21 2018

#linuxcnc-devel Calendar

01:09 AM IchGucksLive: hi i found a small hint
01:12 AM IchGucksLive: at homing it does not respect the rapid override
01:12 AM IchGucksLive: so if i set this to 50 % it shoudent speed up full
01:12 AM IchGucksLive: i know there is a Home_finel_vel
01:12 AM IchGucksLive: but it is not inside stepconf
01:12 AM IchGucksLive: woudt be good if someone cares on this
08:22 AM jepler: homing is not a rapid
08:22 AM jepler: homing does not take place in cartesian coordinates in general
08:51 AM cradek: long ago, homing was affected by feed override, but that was a bug and it was fixed
09:00 AM skunkworks: but I need to machine things while homing!
09:05 AM pcw_home: I suspect 0 spindle speed machining is done fairly often while homing
09:17 AM rene-dev_: https://xkcd.com/1172/
09:56 AM IchGucksLive: hi i did read the logs you answerd about homing ad rapid's
09:57 AM IchGucksLive: the final move is as from stepconf not insert in the config so the move from Referenzofset to Ref point is a Rapid
09:57 AM IchGucksLive: i use only one pin for all 9 Switches
09:58 AM IchGucksLive: and it homes against then latch same direction and then moves to given HOME position WITH MAX Speed and max acceleration
09:58 AM IchGucksLive: on Stepconf first config
09:59 AM IchGucksLive: BETTER TO GET the Home_final_Vel to the same as homing by far not the max mashine speeed
09:59 AM seb_kuzminsky: IchGucksLive: are you saying StepConf should have a way to specify HOME_FINAL_VEL, but it currently doesn't?
10:00 AM IchGucksLive: ofcause this is a way to do it
10:00 AM IchGucksLive: stepconf is basic settings so it mussent
10:01 AM IchGucksLive: better the system takes care
10:02 AM IchGucksLive: IF not HOME_FINAL_VEL then use HOME_SEARCH_VEL
10:02 AM seb_kuzminsky: do you want there to be a different way to specify the velocity of the final move during homing, other than HOME_FINAL_VEL?
10:03 AM IchGucksLive: no
10:04 AM IchGucksLive: HOME_FINAL_VEL is not set in base INI
10:04 AM IchGucksLive: but it is in USE during homing
10:04 AM seb_kuzminsky: do you mean, HOME_FINAL_VEL is not set in the ini written by Stepconf?
10:04 AM IchGucksLive: so where does the number used come from
10:04 AM IchGucksLive: YES
10:04 AM seb_kuzminsky: http://linuxcnc.org/docs/html/config/ini-homing.html#_home_final_vel
10:04 AM seb_kuzminsky: "If the HOME_FINAL_VEL is missing from the ini file, then the maximum joint speed is used to make this move."
10:05 AM IchGucksLive: If the HOME_FINAL_VEL is missing from the ini file, then the maximum joint speed is used
10:05 AM IchGucksLive: WHY there is a SEARCH VEL
10:06 AM seb_kuzminsky: HOME_SEARCH_VEL is used for something different
10:06 AM seb_kuzminsky: if you want to control the velocity of the final move, use HOME_FINAL_VEL
10:06 AM IchGucksLive: Usingthe max as a standard is not the wright way to do it
10:06 AM IchGucksLive: I Personly kow it
10:07 AM IchGucksLive: and ofcause i use it in all my mashines
10:07 AM seb_kuzminsky: the home final move takes place after the joint has been homed, so it seems reasonable to me to default to the configured max velocity of the joint (while allowing the user to reduce it, using HOME_FINAL_VEL)
10:08 AM IchGucksLive: Setting up a VMC in pncconf and try homing it will then speed at maybe 700ipm to the point
10:08 AM seb_kuzminsky: nice!
10:08 AM IchGucksLive: the outer pich most switch are located are not offen used
10:09 AM IchGucksLive: seb_kuzminsky, Thank you for all your work you are on this ones moew a great gift
10:09 AM IchGucksLive: as we now loosing ground with nearly lightspeed to arduino
10:10 AM IchGucksLive: i personly feel the effects
10:10 AM IchGucksLive: more and more schools going to the system
10:11 AM IchGucksLive: ok im off to beergarden its still sunny outside
10:11 AM seb_kuzminsky: enjoy!
12:52 PM KimK: There are things that can go wrong during homing, so my choice would be to limit the machine to 25% or so of its maximum speed until homing is complete and it has arrived at the witness mark, or at least its "usual position" (so it doesn't mistakenly slam into a hard stop at full speed). So I'm curious, why was the ability to limit speed during homing viewed as a bug? It seems like a handy tool for slowing/pausing during the
12:52 PM KimK: switch approach, backoff, etc.
12:58 PM seb_kuzminsky: http://linuxgizmos.com/skylake-module-aces-osadls-real-time-linux-tests/
12:59 PM seb_kuzminsky: KimK: but we do have that ability. all stages of homing are speed-limited based on configuration values in the ini file
12:59 PM seb_kuzminsky: oh, you mean dynamically changing the speed limit during homing?
01:00 PM KimK: Yes, but I mean why was speed-limiting on the fly from the front panel viewed as a bug?
01:00 PM KimK: Yes, exactly.
01:03 PM KimK: So that if some kind of trouble was suspected, that you could slow things down (slower than normal) to see what's happening. (Flakey switch, loose coupling, screws loose, whatever.)
01:04 PM KimK: Especially slowing down during the reversals.
01:11 PM seb_kuzminsky: i'm not sure of the exact reason it was removed
01:11 PM KimK: I'll bet it was because someone couldn't home with his max vel knob set to zero, and complained? (Maybe it was Spacebar-Heater-Guy from the XKCD above?)
01:13 PM seb_kuzminsky: i think cradek was talking about this commit? https://github.com/LinuxCNC/linuxcnc/commit/50b1fbe13fd265513551d16649c0f59db14476d7
01:14 PM seb_kuzminsky: that says, feed override and rapid override don't apply when Motion is in Free mode (used for joint jogging & homing)
01:14 PM seb_kuzminsky: though the commit message only talks about jogging
01:41 PM cradek: my feeling is that on some machines you'd get a different home if you home slower than normal
03:17 PM pcw_mesa: Is this a LinuxCNC TP bug:
03:17 PM pcw_mesa: https://www.cnczone.com/forums/tormach-pathpilot-/354884-tormach.html
03:17 PM pcw_mesa: I can re-create it on 2.7 and master with the gcode snippet and its not fixable with any reasonable G64 settings
03:21 PM seb_kuzminsky: the snippet in comment #6?
03:22 PM seb_kuzminsky: or the second snippet in #3?
03:25 PM pcw_mesa: #6 is the one I tried
03:26 PM pcw_mesa: unless I've missed something wrong with the gcode
03:32 PM seb_kuzminsky: i don't see any unusual motion with the gcode from #6 (i commented out n1710 and appended m2)
03:32 PM seb_kuzminsky: i also tried adding g64 to the top, and pushing feed override to 1000%
03:33 PM seb_kuzminsky: this is all with sim/axis
03:33 PM pcw_mesa: I get a slowdown at the line-arc junctions
03:33 PM seb_kuzminsky: what config are you using?
03:33 PM pcw_mesa: a real stepgen config
03:34 PM pcw_mesa: even with G64 (nothing) it slows down
03:35 PM pcw_mesa: are you plotting velocity?
03:36 PM seb_kuzminsky: yes, i see it now
03:38 PM seb_kuzminsky: http://highlab.com/~seb/linuxcnc/tp-slowdown.png
03:45 PM seb_kuzminsky: in only happens on the first arc, not the second: http://highlab.com/~seb/linuxcnc/tp-slowdown-2.png
03:48 PM pcw_mesa: ISTR this bug or something like it may have been reported before on the forum. Maybe some number of segments must be in the queue before the TP blending works
03:49 PM pcw_mesa: i wonder is splitting the first G1l move into a few segments would fix it
03:50 PM pcw_mesa: wonder if
03:51 PM seb_kuzminsky: i added a bunch of back-and-forth g1 moves before n1760, no change
03:52 PM pcw_mesa: the initial G0 Z moves are a bit odd
03:53 PM seb_kuzminsky: cam...
04:06 PM seb_kuzminsky: i wonder if it's always the first arc in any program?
04:09 PM pcw_mesa: maybe I guess you could clone/double the file and see if its first arc or first arc after G0
04:16 PM seb_kuzminsky: i played with it a bit and i don't see a pattern there yet
04:16 PM seb_kuzminsky: it makes sense that entry to the first arc and exit from the second arc should slow down, since there's a sharp corner there
04:16 PM seb_kuzminsky: but exit from the first arc looks just about tangential, no need for velocity to drop to 0 there
04:17 PM pcw_mesa: Yeah especially with reasonable G64 settings
04:35 PM seb_kuzminsky: i remember years ago cradek noted that we have terrible test coverage of Motion's output...
04:40 PM pcw_mesa: This does seem like a fairly rare bug or there would a lot more complaints
04:43 PM seb_kuzminsky: i was not able to repro the bug on the cds.ngc program
04:48 PM pcw_mesa: I think I tried turning off the new TP and the bug remained
05:18 PM andypugh: Should someone post a link to the bug tracker on the cnczone thread?
05:54 PM seb_kuzminsky: they're running Path Pilot, i don't know what the upgrade path from there to here would look like for them
05:54 PM seb_kuzminsky: so i don't know if they care if we fix it in linuxcnc
05:54 PM seb_kuzminsky: that said, i guess i don't mind if someone links our bug to their thread