#linuxcnc-devel Logs

Feb 09 2018

#linuxcnc-devel Calendar

07:55 AM skunkworks: seb_kuzminsky, it does go into power lower current after a few seconds - but comes right back out normally. it is only the first jog
07:57 AM skunkworks: or maybe movement.
07:57 AM skunkworks: I will goof around with it more
02:26 PM skunkworks: I guess it has to be something with the drive. I just don't know what.
02:26 PM skunkworks: the drive stays on through the restart of linuxcnc and it only has step/dir connected to it.
02:27 PM skunkworks: so I don't know if it gets in an odd state from twiddling of the dir/step pins on the reboot of linuxcnc
02:27 PM skunkworks: hmm hold on
02:29 PM seb_kuzminsky: there should be no wiggle in step or dir when restarting linuxcnc
02:29 PM skunkworks: ok - if I unplug the step/dir cable from the drive - restart linuxcnc - replug the step/dir plug back in I don't get the cluck
02:30 PM seb_kuzminsky: only when commanding motion
02:30 PM seb_kuzminsky: hmm, that's odd
02:30 PM skunkworks: I know - right?
02:30 PM seb_kuzminsky: are step & dir coming from a mesa fpga? if so the output pins will all go high when linuxcnc restarts
02:31 PM skunkworks: 7i76 daughter board
02:31 PM skunkworks: the drive must get into an odd state when the board comes back up
02:31 PM skunkworks: It hasn't stalled - but it sure sounds yecky
02:32 PM KimK: Some stepper drives drop their motor current down to a lower level (position holding) after a few seconds of no-motion, could that be it?
02:32 PM skunkworks: it isn't the mesa stepgen - I actually ran it back into the encoder counter - scopes fine.
02:32 PM KimK: Maybe it
02:32 PM skunkworks: KimK, this one does - but doesn't effect motion. it is only the first jog after a restart of linuxcnc that clunks
02:33 PM seb_kuzminsky: it'd be interesting to see a scope trace of step & dir during a restart and a jog
02:33 PM seb_kuzminsky: probably a physical scope connected at the drive
02:34 PM KimK: first jog in either direction, or just one or the other?
02:34 PM skunkworks: either I am pretty sure
02:34 PM skunkworks: let me test
02:34 PM skunkworks: let me test
02:34 PM seb_kuzminsky: those were both in the same direction
02:34 PM seb_kuzminsky: i think you meant tset em tel
02:35 PM skunkworks: both ways
02:35 PM skunkworks: both ways
02:35 PM skunkworks: focus! stop that!
02:37 PM KimK: Let *me* test. No, let *me* test. No, allow *me*. No, I insist. (etc.,etc.)
02:37 PM skunkworks: I don't have a cool digital scope other than linuxcnc
02:38 PM skunkworks: I guess I could setup another linuxcnc to read the outputs...
02:38 PM skunkworks: ;)
02:40 PM pcw_mesa: You might check that the idle step state ( after linuxcnc exits ) is the drives idea of idle
02:41 PM skunkworks: http://www.leadshine.com/UploadFile/Down/MA860Hm.pdf
02:41 PM skunkworks: pretty basic drive. The enable isn't hooked up (and I don't even know if the enable is accessible...)
02:41 PM skunkworks: on this drive (custom hardware)
02:42 PM skunkworks: having an enable would probably fix this problem (disable drive when linuxcnc isn't on)
02:43 PM pcw_mesa: choosing the other step output (or swapping them if you are driving differentially ) and reversing the step polarity in hal may help
02:44 PM skunkworks: I swapped (these are + common) step - dir - to dir + step +. No effect
02:45 PM skunkworks: I could try up/down I suppose
02:46 PM skunkworks: it certainly doesn't do it when you unplug the step/dir plug on linuxcnc restart - just odd
02:47 PM pcw_mesa: maybe some kind of reset happens to their microstep counter when something odd happens with step or dir
02:48 PM skunkworks: i suppose when linuxcnc restarts step/dir changes in lock step
02:48 PM skunkworks: no timing between them. maybe that screws up the drive
02:49 PM pcw_mesa: yes though that should be avoidable by choosing the correct step output _AND_ changing the step polarity
02:49 PM pcw_mesa:
02:51 PM pcw_mesa: though the driver/hal startup can generate glitches here
02:52 PM skunkworks: having a working enable would fix this issue
02:52 PM skunkworks: most likely
02:52 PM pcw_mesa: I suspect something causes the microstep counter to reset
02:53 PM skunkworks: most likely
02:53 PM skunkworks: heh
03:29 PM skunkworks: I bet it is because they both go low/high or high/low at the same time. when I plug unplug - replug back in - I don't plug it in perfectly
03:29 PM skunkworks: and it doesn't exhibit the behavior..
03:36 PM skunkworks: stupid question - how do you change the step polarity?
03:38 PM andypugh: invert the output pin
03:38 PM seb_kuzminsky: using .invert-output on the gpio?
03:40 PM pcw_mesa: you can use invert-output on the step pin if using master
03:43 PM skunkworks: not using master - so i just have to invert IO Pin 001 (P2-14)
03:45 PM pcw_mesa: that makes it active low at the FPGA interface so should idle and WD bite to idle
03:46 PM pcw_mesa: that is, step should not change state when the WD bites
03:49 PM skunkworks: log\
04:06 PM skunkworks: ok - hooking the step/dir to + and inverting the step input seems to have fixed it
04:07 PM pcw_mesa: should minimize the startup/shutdown warbeling
04:07 PM skunkworks: (I think so the step stays low when powering off
04:07 PM skunkworks: and powering up