#linuxcnc-devel | Logs for 2014-11-27

[01:05:45] <seb_kuzminsky> jthornton_: here's what i'm thinking for the Getting Started guide: http://buildbot.linuxcnc.org/doc/scratch/v2.7.0~pre2~seb-2.7-docs-reorg~c6f77d8/html/
[08:22:35] <jepler> seb_kuzminsky: you might want to change Ubuntu mentions to Debian mentions since that's the OS we're providing current install media for
[08:23:18] <jepler> I think zsync is the preferred download method now
[08:24:04] <jepler> > For configurations that have customized emc.nml, a change similar to this one is required.
[08:24:09] <jepler> probably "this one" used to be a hyperlink
[08:24:23] <jepler> >
[08:24:23] <jepler> Download the firmware images as tar files from http://emergent.unpy.net/01267622561 and install them manually
[08:24:31] <jepler> probably want to point at the right directory on wlo
[08:24:46] <jepler> though .. I dunno what firmware version(s) will work with old linuxcnc like that sooo
[08:25:27] <jepler> > Do not attempt run LinuxCNC while the latency test is running.
[08:25:30] <jepler> ... or vice-versa
[10:24:49] <wilstonoreo> hey guys
[10:25:02] <wilstonoreo> i have a question concerning the traj planner
[10:25:57] <wilstonoreo> first if there has anything changed since version 2.6.0
[10:26:32] <wilstonoreo> because in my current setup, the gcode commands g61 and g64 do not seem to have any effect
[10:26:52] <wilstonoreo> i'm using linuxcnc 2.70 together with a mesa 5i25
[10:27:48] <wilstonoreo> when lots of tiny segments are in the gcode, the machine is driving slower (even if the segments are on a straight line)
[10:28:50] <wilstonoreo> which i thought path planner is smart enough to handle
[10:30:41] <archivist> it should in 2.7
[10:30:47] <wilstonoreo> ok
[10:32:08] <wilstonoreo> so assuming a straight line with several g1 commands is traversed and consists of periods of longer segments and shorter ones (my current test setup).
[10:32:09] <wilstonoreo> In 2.70, the feed rate should be constant then?
[10:32:17] <archivist> the new planner announcement http://comments.gmane.org/gmane.linux.distributions.emc.user/51783
[10:33:57] <archivist> in that thread there is mention of the settings available
[10:34:57] <pcw_home> The new TP (that is turned on by default) in 2.7 should be considerably faster at gcode with lots of short segments
[10:35:16] <wilstonoreo> ah ok, this arc_blend stuff
[10:35:29] <wilstonoreo> the source is in src/emc/tp/tp.c ?
[10:36:34] <pcw_home> I would presume so
[10:37:34] <wilstonoreo> other question, what is advantage of using xenomai instead of rtai?
[10:39:35] <pcw_home> LinuxCNC currently doesn't support Xenomai (Machinekit does)
[10:40:02] <pcw_home> LinuxCNC currently supports Preemt-RT and RTAI
[10:40:21] <wilstonoreo> which one is better?
[10:40:32] <pcw_home> all are better
[10:40:42] <wilstonoreo> Preemt-RT or RTAI?
[10:40:45] <pcw_home> (at different things)
[10:41:28] <wilstonoreo> aha
[10:41:30] <wilstonoreo> why would i use xenomai, for instance
[10:41:52] <pcw_home> latency wise its roughly RTAI, Xenomai, Preemt-RT in order of worsening latency
[10:43:11] <pcw_home> so for software step generation on a PC RTAI is best
[10:43:31] <wilstonoreo> we do hardware step gen
[10:43:32] <Tom_L> does it really matter so much with your boards?
[10:44:19] <pcw_home> No its not a big issue unless you have software stepgens or you need really high servo thread rates
[10:44:50] <Tom_L> i wish there was a latency chart somewhere that showed levels of good and bad
[10:45:26] <Tom_L> and potentially max soft and hard step rates
[10:45:34] <pcw_home> also Xenomai (and Preemt-RT) have support for current Arm CPUs, RTAI does not
[10:45:45] <wilstonoreo> because 2.70 doesnt work with our boards since i have a xenomai kernel. We used machine kit before, but it doesnt not seem to have the new path planner.
[10:45:45] <wilstonoreo> So basically, i setup a debian with Preemt-RT, check out 2.70 and compile?
[10:45:48] <wilstonoreo> we work on x86
[10:46:08] <pcw_home> MK does have the new TP
[10:46:23] <wilstonoreo> ok, then why it doesnt work with my setup?
[10:46:43] <pcw_home> Not sure if its enabled by default like it is in linuxcnc
[10:46:44] <wilstonoreo> one sec, i take a look at which branch i'm currently
[10:47:11] <wilstonoreo> on machinekit
[10:48:27] <pcw_home> you may need to set ini.traj_arc_blend_enable = true
[10:48:53] <wilstonoreo> in the hal file?
[10:48:54] <Tom_itx> pcw_home, i got those chips off the boards if i could get with you some time next week... would be appreciated.
[10:49:34] <pcw_home> OK I'll make a bag-o-chips next week to mail to you
[10:49:51] <Tom_itx> lemme know the damages...
[10:49:58] <pcw_home> you can setp it in the hal file
[10:50:34] <wilstonoreo> so i have the arc_blend stuff in my inifile but nothing changed
[10:50:36] <wilstonoreo> s
[10:50:46] <wilstonoreo> do i need to recompile it?
[10:51:12] <pcw_home> Wouldn't think so
[10:52:19] <wilstonoreo> so when a preemt-rt is sufficient and i only need the new traj planner by default, linuxcnc 2.70 should preferable, shouldnt it?
[10:54:52] <pcw_home> I guess it depends on what you are doing, is debian iso has both kernels available so you can try both
[10:56:16] <pcw_home> if i run the linuxcnc splash screen (1200 IPM) it runs is ~12 seconds with the CAB off and ~8 seconds with CAB on
[10:56:31] <wilstonoreo> cab?
[10:57:02] <wilstonoreo> i'm running a 3d printer with it
[10:57:19] <pcw_home> Circualar Arc Blending (the new TP)
[10:57:44] <pcw_home> circular
[10:58:24] <wilstonoreo> yeah, i saw it, played with the settings in machine kit, but i guess its not activated
[10:58:51] <wilstonoreo> so then i will test it with good old new linuxcnc 2.7.0
[10:59:36] <wilstonoreo> will try that, come back to you guys later maybe, thanks a lot for now!
[10:59:39] <pcw_home> and g64 works as usual
[11:00:18] <skunkworks> I pretty sure the new tp is activated by default in linuxcnc (2.7+)
[11:00:50] <Tom_itx> pcw_home, could the 7i43 be powered without U9 to test for connectivity?
[11:01:25] <wilstonoreo> ok, how can i see (without printing anything) if the new tp is activated?
[11:02:05] <skunkworks> wilstonoreo: are you using the A axis for your extruder?
[11:02:16] <wilstonoreo> yeah exactly
[11:03:19] <wilstonoreo> does the 4th axis affect the trajectory?
[11:03:31] <skunkworks> you won't see any improvement in speed... the new tp only improves XYZ moves... if you add an A movement it falls back to the classic blending
[11:04:20] <wilstonoreo> is there a workaround for this?
[11:04:29] <wilstonoreo> thats pretty odd to be honest
[11:04:37] <skunkworks> switch to velocity extruding - (so it isn't an axis)
[11:05:08] <wilstonoreo> ok, i read about that already... not so easy to set up it seems
[11:09:08] <pcw_home> Tom_itx: Im not sure the power will be enabled without U9
[11:09:08] <wilstonoreo> https://github.com/luminize/machinekit/tree/vel-control-extruding
[11:09:18] <Tom_itx> ok
[11:09:35] <Tom_itx> i was just gonna check the 3.3v reg and see if the pc could see the board
[11:09:53] <Tom_itx> i'll just wait
[11:12:01] <pcw_home> I will send a couple FAN2001s in case they got damaged
[11:12:39] <wilstonoreo> but velocity extrusion doesnt work with two extruders...
[11:13:31] <Tom_itx> i think i'll order a 7i90 7i47 soon just to have as a fallback anyway
[11:13:51] <Tom_itx> do you have those screwterminal power rails on the store page?
[11:14:01] <Tom_itx> i'd like to get one of those too
[11:14:47] <pcw_home> the common block? not sure what the stock situation is on those (we are behind quite a bit)
[11:14:54] <Tom_itx> anything else i'll need for the 7i90?
[11:15:00] <Tom_itx> same d25?
[11:15:23] <Tom_itx> i should get some din rail mounts
[11:15:30] <Tom_itx> do you guys have the rails themselves?
[11:15:43] <Tom_itx> i don't think i've got enough here
[11:16:33] <Tom_itx> i can check back later on all that
[11:18:01] <cradek> > Once I removed the oilers and drilled the holes deep enough to reach the bearings I had a way to inject insufficient quantities of the incorrect lubricant.
[11:18:32] <Tom_itx> stuff it with felt?
[11:19:35] <pcw_home> why would velocity extrusion not work with any number of extruders?
[11:20:32] <wilstonoreo> should be possible with a t command shouldn't it?
[12:12:52] <seb_kuzminsky> jepler: good feedback, thanks
[12:13:14] <seb_kuzminsky> i've not really started fixing up the contents yet, but i noted those items for when i do
[12:13:39] <seb_kuzminsky> so far the thing i've done is reorganized the getting started docs:
[12:15:45] <seb_kuzminsky> whoops
[12:16:21] <wilstonoreo> so the new trajectory planner works for me, but as you guys only without the 4th axis
[12:16:51] <wilstonoreo> just wrote a little python script which generates a test gcode with a straight line and different segment sizes
[12:18:03] <seb_kuzminsky> in 2.7.0~pre2 we had all the sections of the getting-started docs as separate html files, generated from separate asciidoc source files
[12:18:35] <seb_kuzminsky> all the getting-started sections were linked (manually) from the top-level index.html
[12:19:28] <seb_kuzminsky> that branch moves all the getting-started source files to a new getting-started/ directory, and makes a top-level Getting Started html document from them, with the same contents as the Getting Started pdf
[12:48:20] <wilstonoreo> so guys, i hacked the traj planner, so it doesnt fallback when abc axes are used
[12:48:23] <wilstonoreo> works nicely
[12:48:31] <wilstonoreo> thx a lot
[13:04:27] <skunkworks> wilstonoreo: could we see what you did?
[13:06:55] <wilstonoreo> sure, one sec
[13:10:07] <wilstonoreo> http://pastebin.com/dHvkq0n1
[13:10:13] <wilstonoreo> quite hacky, i know
[13:31:27] <wilstonoreo> oh sorry forget, its all in src/emc/tp/tp.c
[13:48:52] <jthornton> seb_kuzminsky, thanks for organizing the docs
[14:41:12] <seb_kuzminsky> jthornton: sure thing! it's just the start so far, since there seems to be a general opinion that it's an improvement i'll go ahead and do the same to the other books (user guide, integrator guide, etc)
[15:26:54] <jthornton> great and don't eat too much today
[15:30:57] <zeeshan|2> do you guys have a good quick reference for someone with some _programming_ experience
[15:31:06] <zeeshan|2> that deals with communication over protocols?
[22:18:47] <skunkworks> logger[psha]:
[22:18:47] <logger[psha]> skunkworks: Log stored at http://psha.org.ru/irc/%23linuxcnc-devel/2014-11-28.html