#linuxcnc-devel Logs

Oct 07 2018

03:16 PM andypugh: I am a bit puzzled by this spindle-at-speed thing
03:18 PM andypugh: The Axis splash G-code in the sim waits for spindle-at-speed, then does the Z feed. Then all motion stops.
03:19 PM hazzy-m: andypugh: Hello! I think I have found several more multispindle bugs, I'm making a more complete VCP panel for the multispindle lathe sim for testing. As yet I can't tell if they are just in the python interface or elsewhere.
03:19 PM andypugh: OK.
03:20 PM andypugh: I am beginning to regret ever starting this
03:21 PM hazzy-m: once the bugs get squished it will be a very valuable feature
03:29 PM rene_dev_: andypugh I was confused, when my drill stuck in the hole
03:30 PM rene_dev_: looking at the code doesnt show anything obvious, but you are waiting for all spindles to start, which should be ok
03:30 PM rene_dev_: I dont understand why it does one move, and then stops
03:31 PM rene_dev_: as far as Im aware, spindle at speed it only checked once, after the s word, and then ignored
03:33 PM andypugh: If you set sim/axis up for two spindles, then use M3 S10000 $1 then sim/axis works..
03:34 PM rene_dev_: https://github.com/LinuxCNC/linuxcnc/blob/c97899d21f9d9496566c2f605d575436b80995b8/src/emc/tp/tp.c#L2584
03:35 PM rene_dev_: the indentation and braces are very misleading
03:35 PM rene_dev_: there are no bracket for the for
03:35 PM rene_dev_: there is one here https://github.com/LinuxCNC/linuxcnc/blob/c97899d21f9d9496566c2f605d575436b80995b8/src/emc/tp/tp.c#L2655
03:48 PM andypugh: Well spotted. Definitely an error, but fixing it doesn’t make things work.
03:50 PM rene_dev_: is emcmotConfig->numSpindles 1?
03:52 PM rene_dev_: im not sure the other for loop does the correct thing.
03:53 PM rene_dev_: my update_ini fis should fix genes problem, but I dont understand any of his latest email...
03:53 PM andypugh: num_spindles definitelu = 1
03:54 PM hazzy-m: Probably not pertinent to what you are looking at there, but the overrides, override_enable and brake settings are sometimes applied to the wrong spindle. Seems like it depends on which spindle was last active
03:54 PM rene_dev_: then it should behave just like before? :D
03:54 PM andypugh: Currently I think that the behaviour is that all at-speed pins need to be set for cutting to happen
03:55 PM rene_dev_: I think they are only checked after S, and ignored after that
03:55 PM rene_dev_: but Im not sure of the behaviour. on some machines the signal might go low, if you start cutting
03:56 PM andypugh: hazzy-m: I think that spindle commands are meant to work on the current “active spindle” though I can see arguments for assuming zero if there is no $ used
03:58 PM hazzy-m: even if a certain spindle is explicitly specified it seems to somtimed be applied to another one
04:20 PM andypugh: Ah, OK. THat might be an off-by one that I already fixed being over-written when I re-factored the commit.
04:21 PM andypugh: But… I am way too tired tonight to get to the bottom of this. I’ll get to it during the week.
04:21 PM hazzy-m: Ok, have you pushed that yet?
04:21 PM andypugh: No, I need to check that it is that.
04:22 PM hazzy-m: Here is a vid showing the strange behavior: https://youtu.be/_Gu_zVesmhc
04:26 PM hazzy-m: That actualy seems like the type of behavior that might be cause by an off by one, I'll wait till you get a chance to look at the more to do any more testing
04:29 PM andypugh: That looks really confusing..
04:30 PM hazzy-m: it is! LOL
05:44 PM rene_dev_: andypugh did you check if the pin value makes its way into tp.c?
05:49 PM andypugh: Yes, it does
05:51 PM andypugh: But there is a wierd handshke I don’t quite get somewhere else: https://github.com/LinuxCNC/linuxcnc/blob/c97899d21f9d9496566c2f605d575436b80995b8/src/emc/motion/command.c#L1067
07:51 PM CMorley: Andy: spindle-at-speed is check before each feed move after a rapid move ... not sure if you were aware.