#linuxcnc-devel | Logs for 2015-12-30

[08:02:52] <jepler> skunkworks: in vismach? It could be right or it could be wrong
[08:03:18] <jepler> skunkworks: In this case, the machine rotates the work. It may be the case that the red backplot lines are staying attached to the work when that rotary moves.
[08:22:04] <skunkworks> oh - duh
[08:22:23] <skunkworks> it was late
[08:28:01] <cradek> that's one of the few setups I've seen where AXIS and vismach previews both look right :-)
[08:28:06] <cradek> Asan knows what he's doing
[08:28:11] <cradek> that's quite a machine model
[08:30:16] <archivist> I would like to know what order he has the geometry set to, it sure looks right
[08:32:19] <cradek> you could [cough cough] ask in the youtube comments
[08:33:06] <archivist> you tube comments are broken in the bowser I am using, I cant even reply to a user in one of my vids
[08:36:07] <archivist> stepgen overshoot what should I fiddle to get it better
[08:37:14] <archivist> didnt skunkworks find something when testing the new trajectory planner
[08:37:33] <cradek> software stepgen or mesa? velocity or position mode?
[08:38:20] <archivist> software, standard settings apart from acceleration which I have slowed down a lot
[08:38:50] <cradek> what is feeding it?
[08:39:21] <archivist> http://www.archivist.info/cnc/wormtest/wormtest.ini
[08:39:46] <cradek> MAX_ACCELERATION = .30
[08:39:47] <cradek> STEPGEN_MAXACCEL = .15
[08:39:53] <cradek> this is wrong
[08:40:21] <archivist> in what way, it is deliberate to be slow
[08:40:52] <cradek> stepgen has less accel allowed than the command has, so it can't possibly track
[08:41:15] <cradek> brb
[08:54:48] <archivist> seems better, I wanted to be slow enough that the output worm stopped without running on much, if any
[08:56:48] <archivist> had a mains dropout this morning so the data grab stopped but got 240 degrees of data http://www.archivist.info/cnc/wormtest/plot_normalised2_2.php
[09:32:05] <cradek> wow 0.1 degrees seems like a lot of error
[09:32:37] <cradek> what worm is this?
[09:36:18] <jepler> and strongly periodic every 30 degrees or so
[09:44:05] <archivist> worm pair out of some instrument, just using it as an example of the periodic errors you can get from gearing, 4 start worm driving a helical
[09:44:28] <archivist> so the 30 degrees is one rotation of the worm
[09:45:10] <cradek> to the eye, does it look good, or bad/worn/bent?
[09:45:11] <archivist> I have seen worse errors in bought dividing heads
[09:45:59] <archivist> by eye it looks and feels fine, I cleaned and oiled before setting up the test
[09:47:04] <archivist> basically low cost worms are going be as bad in a lot of cases from my measurements when back at the clockworks
[09:47:53] <cradek> that's interesting
[09:47:59] <archivist> this is where linuxcnc is a win for diagnosis
[09:49:32] <archivist> the jig http://www.collection.archivist.info/searchv13.php?searchstr=wormtest
[09:52:40] <cradek> neato
[09:52:46] <cradek> is that a super deluxe encoder?
[09:53:37] <archivist> which one, the left is 1000 line the blue is 5000
[09:55:08] <archivist> will be using the first one to debunk microstep "accuracy" then first to second to shoe the worm with stepper out of the equation
[09:55:17] <archivist> show
[09:55:49] <cradek> aha
[09:56:45] <archivist> just watching it and the encoder numbers you can see a commanded step and no movement
[09:58:07] <archivist> must do some averaging/correlation to separate the individual error sources
[10:19:32] <cradek> jepler: www-data (the user) on forum is getting DMARC fallout bounces in email
[10:19:42] <cradek> might want to devnull that - not sure
[10:27:01] <jepler> hmmm thanks
[10:27:25] <jepler> I went ahead and did that
[10:27:46] <cradek> yeah can't imagine they're any use
[12:41:20] <KGB-linuxcnc> 03Sebastian Kuzminsky 052.7 0303f7b 06linuxcnc 10docs/src/getting-started/updating-linuxcnc.txt docs: restore huanyang vfd info * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=0303f7b
[13:27:16] <KimK_laptop> seb_kuzminsky: Hi Seb, I'm trying to help test the 8.2/RTAI, still installing. I got as far as dpkg-check-dependencies(sp?), I have tcl/tk 8.6 (and could pick 8.5) but it wants 8.4; I have something6 and it wants something, other similar problems. Any advice? I'm away from that PC now, but will be there this evening.
[13:28:43] <jepler> which "it" wants 8.4?
[13:28:57] <jepler> dpkg-checkbuilddeps executed in a linuxcnc source tree?
[13:29:08] <KimK_laptop> the dpkg dependency checker, yes
[13:29:35] <jepler> that implies to me that you do not have debian/control generated correctly. When debian/configure generates debian/control, it is supposed to name the correct version of Tcl and Tk
[13:29:44] <jepler> for Debian 8.x AKA Debian Jessie, that is 8.6.
[13:29:57] <jepler> Are you working from a git clone or from something else? How did you invoke debian/configure?
[13:31:11] <KimK_laptop> OK, when I return, I'll go back to the beginning of the directions. Git clone, yes, what was it, rtai-5 ?
[13:31:23] <seb_kuzminsky> KimK_laptop: in addition to jepler's suggestions, make sure you're on a recent version of 2.7 or master (earlier versions don't support the jessie rtai)
[13:31:39] <jepler> the stuff that was in the "rtai-5" branch has been merged into the 2.7 branch, and from there into master
[13:31:43] <seb_kuzminsky> the rtai-5 branch has been merged into 2.7 (and up into master)
[13:31:44] <jepler> so using the rtai-5 branch is not necessary anymore
[13:31:55] <KimK_laptop> OK, I was going by the directions.
[13:32:18] <jepler> echo "unknown distribution: $DISTRIB_NAME"
[13:32:31] <jepler> if debian/configure prints "unknown distribution" then we need to know what DISTRIB_NAME is for you
[13:32:51] <jepler> but we expect it to reach an earlier clause that applies to Debian-8.*
[13:33:05] <seb_kuzminsky> KimK_laptop: i see, the instructions in the blog entry on the linuxcnc.org front page
[13:33:14] <seb_kuzminsky> yes, that's outdated now, i should update it
[13:34:13] <KimK_laptop> The jessie install distrib? The directions weren't too specific, but I picked i386 and xfce, trying to avoid trouble.
[13:34:23] <jepler> in fact you should probably just pastebin whatever debian/configure is printing when you run it
[13:34:23] <KimK_laptop> Yes, the front page directions.
[13:37:03] <KimK_laptop> OK, will pastebin tonight. The dependency checker was down to about 4-5 problems, though, so, better.
[13:41:11] <KimK_laptop> Thanks, gents.
[13:47:15] <seb_kuzminsky> whoops, i forgot to merge the rta-5 support from 2.7 into master
[13:47:20] <seb_kuzminsky> *rtai-5
[13:47:29] <KGB-linuxcnc> 03Sebastian Kuzminsky 05master 0e089af 06linuxcnc 10(5 files in 4 dirs) Merge remote-tracking branch 'origin/2.7' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=0e089af
[14:42:25] <CaptHindsight> https://bits.debian.org/2015/12/mourning-ian-murdock.html
[14:42:43] <jepler> I saw that news as well. :(
[14:42:58] <CaptHindsight> young guy
[14:52:19] <jepler> somehow I had former DPL Ian Jackson merged with Debian founder Ian Murdock in my head
[14:54:47] <mozmck> http://sdtimes.com/ian-murdock-has-died/ - The tweets linked are interesting...
[15:07:20] <cradek> I hope we learn the rest of this story, but I bet we won't.
[17:45:58] <KGB-wlo> push to master branch: http://linuxcnc.org/
[17:47:16] <KGB-wlo> push to master branch: http://linuxcnc.org/
[20:23:21] <jepler> https://twitter.com/dallandrummond/status/678972096156983296
[20:26:11] <jepler> http://ch00ftech.com/2015/12/18/electrolytic-capacitors-and-preserving-a-family-heirloom/
[20:47:31] <PCW> I should fix my Friden 132
[20:53:24] <jepler> cradek used to have something that looked vaguely like that in his basement. I wonder what it is I'm remembering.
[20:57:02] <seb_kuzminsky> looks like the computer that runs his lathe
[20:57:07] <seb_kuzminsky> ooo burn
[20:58:20] <seb_kuzminsky> i find i prefer to communicate by writing blog posts on jekyll, which has no comment system, over the mailing list and the forum, i wonder what that says about me
[20:59:57] <andypugh> Isn’t it more like a diary of nobody else reads it?
[21:00:49] <seb_kuzminsky> heh yeah
[21:35:22] <tperry> @linuxcnc-devel
[21:35:27] <tperry> #linuxcnc-devel
[21:43:09] <tperry74> I was looking for the rtai-5 branch to try out jessie, but it does not seem that that branch exists, any pointers?
[22:25:27] <andypugh> Folk with random questios that make no sense need to hang around longer.
[22:35:10] <seb_kuzminsky> it's a good question, and if they'd hung around they'd have learned the answer
[23:56:04] <KimK> Speaking of the now-merged-into-master-and-2.7 rtai-5 branch, things are going much better on master this time, I got "successfully configured for 'Debian-8.2'-'3.16.0-9-rtai-686-pae'..", and now have a make going that looks promising. Thanks again jepler and seb_kuzminsky . More good news shortly, I hope.
[23:57:03] <seb_kuzminsky> great!