#linuxcnc-devel Logs

May 27 2020

#linuxcnc-devel Calendar

07:02 AM cerna: The deb-jessie-rtpreempt-binary-i386 has pending one build but is actually idle: http://buildbot.linuxcnc.org/buildbot/json/builders/4019.deb-jessie-rtpreempt-i386/?as_text=1 - connected to 4343252d223d48aa96accc8b0787922bf2042f12 (for about 4 hours now).
08:17 AM jepler: It's not clear to me why that is "waiting" either. seb_kuzminsky ?
09:32 AM cerna: Well, at least now I know that I will need some state preservation. Because my phone has been beeping for hours now.
12:30 PM cerna: I changed the script to not error out on first hit (few seconds more has no meaning) and discovered that 4020.deb-jessie-rtpreempt-amd64, 4040.deb-buster-rtpreempt-i386, 4041.deb-buster-rtpreempt-amd64, 4042.deb-buster-rtpreempt-rpi4 are also problematic.
02:50 PM Cloudio: Hi, I have found LinkedIn profile of somebody https://cz.linkedin.com/in/jan-mr%C3%A1zek-ba74a364 claiming to implement trajectory planner with acceleration and jerk limitation for LinuxCNC, however I cannot find anything about this on the web or forum. Does LCNC have a jerk limitation?
03:01 PM Cloudio_ is now known as Cloudio
04:11 PM andypugh: Cloudio: No, not at the moment. But Ariadobo dod make it work for them (with some limitations)
04:13 PM andypugh: https://github.com/araisrobo/linuxcnc
04:34 PM rmu|w: https://blog.honzamrazek.cz/wp-content/uploads/2016/07/Text.pdf
04:36 PM Cloudio: andypugh: Thanks andypugh! I hate it when people lie on their Likedin. And I am not even HR monkey.
04:37 PM andypugh: I have no reason to believe that he didn’t do it.
04:38 PM rmu|w: Cloudio: I don't read czech, but have a look at that pdf i just posted a link to if you do.
04:40 PM andypugh: Searching my email archive, Jan Mrázek is credited with some work on the 2.7.5 release, so is a real person, with LinuxCNC involvement.
04:42 PM andypugh: Though it wasn’t a major contribution: https://github.com/LinuxCNC/linuxcnc/commit/9daebb8e6273b5ea2f5558a3ce18e034d848b9a6
04:42 PM andypugh: rmu on the forum has also done some work on this.
04:43 PM rmu|w: that is the same rmu as me
04:43 PM andypugh: The name has popped up 5 times on Hackaday. too.
04:43 PM andypugh: Doh!
04:45 PM pcw_home: rmu|w I'm thinking more and more that that crazy move problem is a race condition as you suggested. it seem harder to rtigger on the RPI4 than my (much faster) X86 setup
04:49 PM rmu|w: pcw_home: calling emcTaskPlanReset seems to make a difference, i didn't have time yet to trace it all out. probably this bug exists for quite some time
04:54 PM pcw_home: Yes adding the reset fixes as far as I can tell.
04:55 PM Cloudio: rmu|w: It looks like a High School competition work about design of new motion controller. Don't take me wrong, I kind of admire him, because I was too lazy and undisciplined at that age to come with something like it, and I am sure he is somewhere else nowadays, but it doesn't solve the problem in LinuxCNC scope.
04:56 PM rmu|w: Cloudio: it looks like the math that is involved is in the paper
04:59 PM andypugh: Yes, it looks like a school project ( https://translate.google.co.uk/translate?sl=cs&tl=en&u=https%3A%2F%2Fcs.wikipedia.org%2Fwiki%2FGymnázium_Kroměř%25C3%25ADž ) is the Czech Wiki page about the school.
05:00 PM andypugh: I am rather impressed at a school pupil getting that much in to LinuxCNC.
05:03 PM rmu|w: the building looks a lot like the school i went to
05:08 PM rmu|w: pcw_home: FYI, this https://github.com/LinuxCNC/linuxcnc/issues/865 is the crazy move github issue
05:13 PM rmu|w: because the "real" close implies task reset, which seems to clear all queues
05:14 PM pcw_home: I don't even know the purpose of lazy_close
05:15 PM rmu|w: also the fix for the on_abort_command motion bug was to call emcTaskPlanClose (in consequence emcTaskPlanReset)
05:15 PM pcw_home: I wonder who is the task Guru, what they would say about adding emcTaskPlanReset();
05:15 PM rmu|w: me neither.
05:49 PM jepler: It is to allow o-word subs from the loaded gcode file to be called from MDI
05:56 PM pcw_mesa_: So the file stays open?
06:07 PM andypugh: This is the bug that is in 2.7 too?
06:13 PM pcw_mesa_: might be, its a bit unpredictable
06:14 PM andypugh: I was thinking of 867 (very different bug, I just remembered that it was in 2.7.15)
06:16 PM andypugh: 865 _might_ be related to the fix made for this: https://github.com/LinuxCNC/linuxcnc/issues/709
06:17 PM andypugh: Is the problem ameanable to a git bisect?
06:17 PM andypugh: ie, is it relatively easy to tell if a specific build has the bug?
06:21 PM pcw_mesa_: On my test machine I have to try to trigger the bug about 5 times before it shows up
06:24 PM andypugh: Well, that’s not too bad. the RTAI bug I am taking a rest from needed 20,000 tests sometimes.
06:24 PM andypugh: Can you check if the problem is in 2.8?
06:24 PM pcw_mesa_: I think rmu is right. it skips forward (say a few hundred lines in my case)
06:25 PM norias: howdy
06:25 PM norias: subscale.io
06:25 PM norias: curious what ya'll think of that
06:26 PM andypugh: It looks like a URL only smaller
06:27 PM pcw_mesa_: I can try 2.8 later
06:27 PM pcw_mesa_: I expect its there
06:27 PM norias: andypugh: good
06:29 PM andypugh: There is a lot more info in the privacy policy than in the front page.
06:30 PM andypugh: (For example, one might want to know where they are, were one applying for a job with them)
06:30 PM norias: yeah
06:30 PM norias: makes sense
06:31 PM norias: that _is_ a long privacy policy
06:31 PM andypugh: (Or a statement like “we only exist on the cloud and welcome applications from as far afield as Pluto)
06:40 PM pcw_mesa_: Hmm are they "on the cloud" or "in the cloud" the former makes me think of Greek gods
06:41 PM norias: oh...
06:41 PM norias: yeah, on the cloud
06:41 PM norias: wait, isn't that more "on mount Olympus" ?
07:04 PM pcw_mesa_: I guess I think of Zeus throwing lightning bolts down from on top of clouds
07:59 PM skunkworks: the little emco hasn't had a noise related estop since I star grounded a few things...
08:00 PM skunkworks: go figure
08:00 PM skunkworks: Plus I moved some i/o from the raw 5i25 plug to the BOB
08:00 PM skunkworks: I had 5 opto 22's - now I am down to 2
08:01 PM skunkworks: those cheap 5 axis bob's are actually pretty decent..
08:02 PM skunkworks: I think the amc amp I am running the spindle motor is saturating it.. ;)
08:03 PM skunkworks: it is rated for 90v and it stops going faster at about 6v going to the +/-10 volts of the amp..
08:03 PM skunkworks: (120vac into the amp)
08:17 PM Tom_L: is it at full speed at 6v?
08:21 PM skunkworks: I don't know.. Have not tach'ed it yet.
08:21 PM skunkworks: I assume so - the belted down speed is over 2400rpm..
08:23 PM skunkworks: (it's highest rpm - is 3000..) And that is at the minimum reduction.
08:24 PM Tom_L: workin over one of the lathes now ehh? givin the green machine a rest...
09:11 PM skunkworks: yes - Mainly wanted to try some fancy non-circular turning :)
09:13 PM skunkworks: looks like the motor is 4000 rpm 95v
09:21 PM Tom_L: huh
09:21 PM Tom_L: dc?
09:23 PM skunkworks: yes
09:23 PM skunkworks: brushed dc
09:24 PM skunkworks: I don't know if that is exactly right. Going to have to tach it myself I guress
09:24 PM skunkworks: guess
09:25 PM Tom_L: sounds pretty close to the sherline motor
09:25 PM Tom_L: it's only ~600w though
09:33 PM skunkworks: this is around that I think
09:34 PM skunkworks: .44kw
09:34 PM skunkworks: dinky
09:34 PM skunkworks: but lots of belt
10:49 PM skunkworks: we were going to measure the voltage going to the motor to see if it keeps going up.. But didn't want to use a 'good' volt meter and couldn't find our analog one.