#emc-devel | Logs for 2008-01-27

[00:01:03] <Roguish> jepler. back. switched cables. gonna try again.
[00:04:17] <Roguish> now the correct motors are running.
[00:08:38] <jepler> I looked around a little more for an explanation of linux ordering of pci devices but I didn't find anything yet that explains it in detail
[00:10:22] <jmkasunich> cradek: around? I have a tru-type-tracer question
[00:10:36] <jmkasunich> are there "stick fonts" that work with TTT, or does it always outline the letters?
[00:11:00] <jmkasunich> seems like stick fonts would retain legibility better at small sizes, and be faster to machine
[00:18:25] <cradek> there are some sticklike fonts, but not many of them
[00:19:43] <jmkasunich> any names come to mind, or should I just load open office and start trying different fonts?
[00:19:53] <cradek> sorry
[00:20:00] <cradek> I have something, one minute
[00:25:35] <jmkasunich> the closest thing I see to a line font is "Baekmuk Gulim", but I bet even it is implemented as an outline
[00:54:51] <fenn> jmkasunich: hershey is a stick font, and lucky you it comes with emc
[00:55:13] <cradek> I think only 0-9, XYZ, minus, dot
[00:56:33] <jmkasunich> fenn: yeah, I'm finding more complete hershey stuff online
[00:56:36] <SWPadnos> Roguish, I think the PCI scan order has no relation to the slot order
[00:57:12] <SWPadnos> I remember sometime in the 2.4 (or possibly 2.2) series, the order changed for some reason, causing many problems with RAID setups (with multiple SCSI adapters), among other things
[00:57:29] <fenn> jmkasunich: there's more fonts in /usr/share/qcad/fonts/
[00:58:02] <fenn> weird format tho i guess
[00:58:27] <cradek> chris@rover:/usr/share/qcad/fonts$ truetype-tracer -f romans.cxf asdf
[00:58:27] <cradek> Fatal error in FT_New_Face: unknown file format (2) at line:255
[00:58:28] <cradek> darn.
[01:37:13] <Roguish> ok, little help please, what does this mean: core_servo.hal:42: Ini variable not found. i'v been thru the core_servo.hal file tooooo many times and cannot see any problem.
[01:37:35] <Roguish> is this line 43 or error 42?
[01:37:40] <jmkasunich> line 42
[01:37:45] <Roguish> line 42
[01:38:02] <Roguish> with or with out comment lines?
[01:38:05] <jmkasunich> it probably has something like "setp <someparam> [AXIS_0]FOO"
[01:38:18] <Roguish> checking again.
[01:38:24] <jmkasunich> the [AXIS_0]FOO references a variable in the ini file
[01:38:47] <jmkasunich> specifically variable FOO in section [AXIS_0]
[01:39:04] <jmkasunich> and that error message is saying that there is no variable FOO in section AXIS_0
[01:39:13] <Roguish> ok, the file does not have FF2. does it need to?
[01:39:43] <jmkasunich> only if you wish to set the FF2 pid gain
[01:40:10] <Roguish> what is FOO?
[01:40:45] <jmkasunich> http://en.wikipedia.org/wiki/Foo
[01:41:00] <jmkasunich> basically a placeholder, for whatever name
[01:41:31] <Roguish> ok, the .ini file only has FF0 and FF1.
[01:42:01] <jmkasunich> you can either delete or comment out that hal file line (if you don't want to use FF2) or add an FF2 line to the ini file
[01:42:19] <jmkasunich> I'd add the line to the ini file, set the variable to zero if you don't want to use it
[01:42:41] <Roguish> and add FF2 to the core_servo?
[01:43:10] <jmkasunich> huh?
[01:43:34] <jmkasunich> core_servo line 42 already has FF2, thats why you are getting the error
[01:43:56] <jmkasunich> because core_servo.hal is asking for something that the ini file doesn't have
[01:44:17] <Roguish> no, there is NO FF2 in the file.
[01:45:04] <jmkasunich> I obviously misunderstood something you said a while back
[01:45:14] <jmkasunich> you are the one who brought up FF2
[01:45:32] <jmkasunich> so I assumed that you found a reference to FF2 on line 42 of core_servo.hal, and that is what caused the error
[01:46:36] <Roguish> is line 42 literal, that is counting all lines including comments. or actual lines skipping comments and blank lines?
[01:46:59] <jmkasunich> all lines
[01:47:10] <Roguish> ah, ok. i'll count.
[01:47:14] <jmkasunich> don't count
[01:47:17] <jmkasunich> open it in an editor
[01:47:37] <jmkasunich> (any half-decent text editor displays the line number, and often the column number, where the cursor is)
[01:48:15] <Roguish> think i found the problem........
[01:48:22] <jmkasunich> in gedit (the ubuntu default editor, there is "line 1 col 1" at the bottom right of the window
[01:48:38] <jmkasunich> actually "ln 1, col 1"
[01:49:53] <Roguish> oh, coool. new error. core_servo.hal:9: exit value: 1
[01:49:55] <Roguish> core_servo.hal:9: insmod failed, returned -1
[01:51:01] <Roguish> i had previously forgot the [AXIS 4] line in the ini file. put it in and now the new error.
[01:51:13] <Roguish> d'oh.
[01:51:14] <jmkasunich> thats odd, usually when you find a new error, its after the one you fixed
[01:52:07] <Roguish> here's the real, new error: insmod: error inserting '/home/noel/emc2-trunk/rtlib/pid.ko': -1 Invalid parameters
[01:52:47] <jmkasunich> paste line 9 of core_servo in here
[01:53:38] <Roguish> # PID module, for three PID loops
[01:53:40] <Roguish> loadrt pid num_chan = 6
[01:54:05] <Roguish> should read 'six' PID loops.
[01:54:08] <jmkasunich> try numchan=6 (no spaces)
[01:54:30] <jmkasunich> it is _extremely_ rare that spelling out a number is the right thing to do ;-)
[01:56:07] <Roguish> same error.
[01:56:46] <jmkasunich> look in dmesg for anything with "pid" in it
[01:56:47] <Roguish> # PID module, for six PID loops
[01:56:48] <Roguish> loadrt pid numchan=6
[01:57:42] <jmkasunich> http://www.linuxcnc.org/docview/html//hal_rtcomps.html#sec:PID
[01:57:50] <jmkasunich> the manual says num_chan
[01:57:56] <jmkasunich> note the underscore
[01:58:55] <jmkasunich> its a good idea to always either check the manual or copy from a known-good file when dealing with picky little details like this
[01:59:00] <Roguish> ok, now i'm gettign somewhere. new error.
[01:59:10] <jmkasunich> computers are stubborn, they want it exactly the way they expect it
[01:59:41] <Roguish> why can't i put spaces in the line: num_chan = 6 ?
[02:00:10] <jmkasunich> because the command processing code splits things at whitespace
[02:00:20] <jmkasunich> num_chan=6 is one thing
[02:00:38] <jmkasunich> num_chan = 6 is three things: "numchan", "=", and "6"
[02:00:47] <Roguish> is that true thoughout the code?
[02:00:58] <jmkasunich> pretty much
[02:01:08] <Roguish> oh pooooooo
[02:01:13] <jmkasunich> certainly for just about all hal component loadrt arguments
[02:01:39] <Roguish> i like spaces. they help me stay organized
[02:01:50] <Roguish> at least as best as i can be
[02:05:15] <Roguish> yay!!!!!! success, sort of. got axis to come up. thanks jmkasunich.
[02:05:35] <jmkasunich> you're welcome
[02:06:31] <Roguish> damn, allllll 6 motor jog!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!1
[02:06:58] <Roguish> this is 2 m5i20 boards in on machine.
[02:07:17] <Roguish> don't think it's been done before. cooooooooooooool
[02:09:08] <Roguish> thank you. thank you, thank you......
[02:20:35] <fenn> * fenn doesnt know whether to applaud or say "you're welcome"
[03:11:11] <cradek> we sorely need a pid tuning document that has scope plots
[03:17:59] <cradek> I think I have a systematic approach that works ok
[03:21:27] <Roguish> jepler: i'm running 'cone.ngc' with 2 m5i20 cards and 5 motors. i think it's running ok.. had to mod the gcode a bit to fit my xy.
[03:21:36] <Roguish> bbl. dinner calls.
[03:22:40] <cradek> I will make these docs when I do my bridgeport if I don't get a chance to work on something else sooner
[03:23:01] <cradek> doubt my little lathe would be a good example. need something bigger.
[03:23:44] <cradek> Roguish: are you working on a 5 axis machine?
[03:44:53] <Roguish> cradek: yes. i have client with a 3 axis gantry machine on FLASHCUT that wants 2 axes added to the spindle. he is aware that we have to change control software and drives.
[03:45:13] <Roguish> i'm doing prep work.
[03:45:54] <SWPadnos> cool
[03:46:05] <SWPadnos> do you know the specs of the flashcut signal generator?
[03:46:08] <cradek> we're still pretty early in the 5 axis work in emc. are you hoping to help with the software too?
[03:46:12] <SWPadnos> (like step rates)
[03:46:44] <Roguish> no, not sure of the step rates, but i do know it definitely limits the machine speed.
[03:47:00] <SWPadnos> interesting. it may be a communication speed issue (block rate)
[03:47:16] <Roguish> i can help in pretty much anyway other than programmin. that i'm not.
[03:54:32] <Roguish> over the next couple of weeks i hope to make a small, very simple B,C axes head.
[03:56:15] <SWPadnos> hmmm. I didn't notice a way to get to this page on linuxcnc.org: http://www.linuxcnc.org/handbook/part2/pidtune.html
[03:56:23] <SWPadnos> I found it by searching for "tuning"
[03:56:53] <jepler> SWPadnos: /handbook must be stuff that is older than the lyx documentation, or at least didn't make it when the docs were put in the emc2 source tree
[03:57:05] <SWPadnos> hmmm. ok
[03:57:29] <SWPadnos> I know there was an article (on the wiki maybe) that had a description of several methods, including ziegler-nichols, but I can't find it now
[03:58:13] <jepler> there's also this: http://www.linuxcnc.org/docview/2.2/html//motion_pid_theory.html
[03:58:27] <jepler> which I think is taken from wikipedia
[03:58:30] <SWPadnos> ok, that's the one
[03:58:37] <SWPadnos> did you search or click links for that?
[03:59:05] <jepler> linuxcnc.org > documentation > current version > all documentation (html) > PID Theory
[03:59:06] <SWPadnos> oh - it's through the documentation links
[03:59:14] <SWPadnos> ok, thanks
[03:59:24] <jepler> but of course I am a bot and have the URL in my associative hash table
[03:59:42] <SWPadnos> hmmm. I'll need to come up with some Turing test questions
[04:58:35] <Roguish> SWPadnos: the 2 m5i20 cards seem to be working. got it set up with 5axiskins and 5 real motors, live even.
[04:58:49] <SWPadnos> cool
[04:59:12] <Roguish> found an interesting param: 5axiskins.pivot-length
[04:59:32] <Roguish> pretty important. it has serious effect.
[04:59:43] <SWPadnos> heh - I'd imagine
[05:01:20] <Roguish> got my 2 7i37t boards and forgot to get the plug in terminal strips. d'oh.
[05:02:06] <Roguish> EVERYONE: thanks for the help tonight. back at it in the morning.
[05:03:12] <SWPadnos> heh
[05:03:23] <SWPadnos> have a good night. see you in the morning :)
[18:32:38] <CIA-21> EMC: 03alex_joni 07TRUNK * 10emc2/src/emc/usr_intf/halui.cc: typo in a comment
[18:33:57] <CIA-21> EMC: 03alex_joni 07TRUNK * 10emc2/docs/src/gui/halui.lyx: mention analog jogging, add some comments
[19:23:36] <alex_joni> I just updated the stats at: http://eneas.juve.ro/~juve/stats/
[19:24:36] <alex_joni> (it's a bit off as it counts dxf lines too.. I think the dxf's commited by fenn aren't marked binary (~140k lines))
[19:47:09] <fenn> heh i sure didnt write 140k lines of code
[20:15:54] <alex_joni> fenn: are you sure?
[20:16:01] <alex_joni> maybe you weren't paying attention :P
[20:56:00] <alex_joni> jepler: around?
[20:56:44] <jepler> alex_joni: no
[20:56:52] <alex_joni> heh.. asquare?
[20:57:07] <SWPadnos> anoctagon
[20:57:13] <alex_joni> found a possibly nicer IRC bot
[20:57:24] <SWPadnos> nicer than jepler?
[20:57:28] <alex_joni> http://supybot.com/
[20:57:34] <alex_joni> SWPadnos: nicer than logger_dev
[20:57:43] <SWPadnos> heh :)
[20:59:51] <SWPadnos> I wonder if it would be able to stay connected any better
[21:00:06] <SWPadnos> since the problems are likely the fault of DH
[21:00:36] <alex_joni> and netsplits
[21:00:42] <alex_joni> which aren't DH's fault
[21:00:50] <SWPadnos> right
[21:01:59] <alex_joni> well.. supybot has a LART plugin
[21:02:01] <jepler> yeah the main feature I'd want is reliably staying connected
[21:02:12] <alex_joni> so that might be _really_ usefull
[21:02:14] <alex_joni> http://supybot.com/documentation/plugins/lart/document_view
[21:05:46] <CIA-21> EMC: 03jepler 07v2_2_branch * 10emc2/debian/emc2.files.in: new file must be included in package
[22:53:56] <skunkworks> lots of activity on the email list. That has to be good. :)
[23:15:07] <Roguish> hey, anybody have any 5 axis gcode? to test run in '5axis' (besides cone.ncg)
[23:16:08] <skunkworks> you could ask dmess...
[23:16:32] <skunkworks> go on over to #cam
[23:16:42] <SWPadnos> cradek may have put the code he used in his video somewhere
[23:16:47] <Roguish> jepler: pivot-length is in mm, isn't it?
[23:17:04] <SWPadnos> should be user units, I'd think (but I don't know for sure)
[23:17:17] <SWPadnos> err - machine units, that is
[23:17:20] <alex_joni> SWPadnos: I was pondering the same thing, but I think not
[23:17:33] <SWPadnos> yeah, I guess it makes sense to define that in one system only
[23:17:34] <alex_joni> it's internal.. and things are mostly mm there.. iirc
[23:17:40] <Roguish> it's got a default in the c file of 250.
[23:17:42] <SWPadnos> since you can have axes with different underlying units (in theory)
[23:17:57] <alex_joni> SWPadnos: only externally (ini)
[23:18:03] <SWPadnos> right
[23:18:31] <Roguish> i've added the cool 5axisgui to my setup.
[23:18:50] <Roguish> and modified cone.gnc to fit my system.
[23:20:12] <Roguish> the sim 5axis gui is definitely in mm as it has dimensions in the hundreds.