#emc-devel | Logs for 2009-01-09

[03:07:57] <jmkasunich> ok, this is wierd
[03:08:10] <jmkasunich> G3 arcs are supposed to be counter-clockwise, right?
[03:08:20] <jmkasunich> I'm watching one go clockwise
[03:09:51] <jepler> yes
[03:09:59] <jepler> I'm sure it's cradek's fault
[03:10:14] <jmkasunich> this is lathe code, XZ plane
[03:10:27] <jmkasunich> working on a simpler test case....
[03:11:30] <jepler> if it can run on sim/lathe I'll give it a shot when you have it
[03:12:26] <jepler> I have a checkout from before the comp stuff was moved to TRUNK, I'll preserve that and also get an up to date tree
[03:12:42] <jmkasunich> I'm running 2.2.8
[03:12:45] <jepler> oh
[03:12:56] <jepler> you didn't mention *that* :-P
[03:13:03] <jmkasunich> G20
[03:13:03] <jmkasunich> G64 P0.0005
[03:13:03] <jmkasunich> G95 F0.010 (set feed-per-rev mode and default feed)
[03:13:03] <jmkasunich> G18
[03:13:03] <jmkasunich> M3S180
[03:13:04] <jmkasunich> G0 X1.0 Z 1.5 (so the next move is visible in preview)
[03:13:06] <jmkasunich> (testing G3 arc direction)
[03:13:08] <jmkasunich> G0 X 1.0 Z 1.2
[03:13:10] <jmkasunich> G3 X 1.0 Z 0.8 I 0.2 K -0.2
[03:13:12] <jmkasunich> G0 X1.2 Z1.0 (move to center of arc)
[03:13:31] <jmkasunich> that is supposed to go from about 1:30 ccw to 10:30
[03:13:37] <jmkasunich> instead it goes cw
[03:15:10] <jmkasunich> I didn't try that in sim, I'm just previewing it on the machine
[03:17:26] <jepler> I'm going to say this and it's going to sound stupid, but here goes
[03:17:40] <jepler> if you load that in a mill configuration, it depends which "side" you look at it whether it's counterclockwise or not
[03:17:48] <jmkasunich> yeah
[03:18:07] <jepler> for the so-called XY view, positive Z is towards the viewer
[03:18:10] <jmkasunich> I was doing hand gymnastics here, transferring coordinate systems from mill to lathe
[03:18:25] <jepler> when I turn positive Y towards me, G3 goes counterclockwise
[03:19:41] <jmkasunich> this is XZ
[03:20:19] <jmkasunich> for a mill, +Y is usually away from the operator, +Z is up, +X is right
[03:21:07] <jmkasunich> for a lathe, +X is toward operator, +Z is right, and +Y is down (I think)
[03:21:45] <jepler> I think that what you just said is like the view presented to the user: +X is towards the bottom of the screen, +Z is right, and +Y is increasing distance into the screen
[03:21:51] <jmkasunich> for the mill, the operator is looking from -Y toward +Y, and the same for lathe
[03:22:36] <SWPadnos> the view presented for lathe mode is a right-hand coordinate plane
[03:22:47] <jmkasunich> lathe and mill are both right handed
[03:22:50] <SWPadnos> which is correct. I don't think that's the problem
[03:22:52] <SWPadnos> exactly
[03:23:14] <jmkasunich> the diff is that for typical milling arcs (XY plane), the view is from +Z towards -Z
[03:23:18] <jepler> I wish cradek is here. I'm sure he always knows what arcs he wants
[03:23:19] <jmkasunich> Z points out of screen
[03:23:35] <jmkasunich> for XZ arcs (lathe or mill), the Y axis points into the screen
[03:24:16] <jmkasunich> well, I'll change the arcs from G3 to G2 and move on with making my parts
[03:24:33] <jmkasunich> it would be interesting to know what the convention is for lathe arcs
[03:51:52] <steves_logging> steves_logging is now known as steve_stallings
[03:54:03] <steve_stallings> jelper - RE: servo drivers using step/direction input, yes there have been many such devices. Back in 1999 Dan Mauch was selling such a device designed by Bill Wainright called the ServoLite. There was also a version done in England and sold by the English dealer for aHHa CNC software.
[03:55:14] <steve_stallings> The old ServoLite web site has been partially recreated here: http://www.rotordesign.com/servolite/step_and_direction.html
[03:56:08] <SWPadnos> high end Yaskawa and Panasonic drives, and presumably others in their class, will also accept step/dir
[03:56:13] <SWPadnos> in addition to other things
[04:06:32] <cradek> jmkasunich: you are right. g3 arcs are only ccw viewed from the positive end of the perpendicular axis
[04:07:05] <cradek> on a lathe in AXIS's view, you're on the negative end of the (nonexistent) Y axis, so the arcs are the other way
[04:07:55] <cradek> or whoever figured that out is right - I didn't read every word
[04:08:37] <cradek> it's not really a lathe problem - G18 arcs are G18 arcs
[04:08:52] <SWPadnos> the plane should be ZX, not XZ
[04:09:05] <SWPadnos> but it makes more sense to change the direction of the perpendicular axis
[04:09:20] <SWPadnos> (the three should be XY YZ and ZX, for rotational symmetry)
[04:09:24] <cradek> on your mill with X right and Z up, you'll see the same "backward" G18 arc if looking from the front
[04:09:41] <cradek> yes I call it ZX but the interp has XZ everywhere
[04:09:46] <SWPadnos> yep
[04:11:42] <cradek> the response about cutter comp is underwhelming. maybe most people never use it.
[04:12:25] <SWPadnos> wait for people like Jon E and Ray to try it out
[04:12:45] <SWPadnos> there should be more interest over the weekend, I bet ;)
[04:14:23] <cradek> at least nobody has tried to argue that it's a bad change
[04:14:30] <cradek> I was a little afraid of that
[04:14:43] <SWPadnos> there's only one possible downside that I can think of
[04:14:59] <SWPadnos> and that's that an incorrect program may run anyway, rather than annoyingly refusing to run
[04:15:22] <SWPadnos> but that's a theoretical possibility, I don't know that it can happen in practice
[04:15:41] <SWPadnos> (because I haven't had a chance to look at the code, and I probably wouldn't understand it if I did)
[04:16:06] <cradek> you can certainly (right now) end up with a gouged part if you feed it the wrong thing
[04:16:27] <SWPadnos> right now as in "with the changes
[04:16:29] <cradek> that was a little less easy to do previously, but certainly was not impossible
[04:16:30] <SWPadnos> "?
[04:16:32] <cradek> yes
[04:16:36] <SWPadnos> ok
[04:16:42] <SWPadnos> so that's the only possible downside I can see
[04:19:50] <cradek> I admit to heavily relying on the preview when writing/proofing comped code
[04:20:37] <cradek> I think I could detect the same kind of gouges as before (the localized ones)
[04:20:38] <SWPadnos> unfortunately, I think a lot of users would prefer to use CAD/CAM software, and expect the code to do what they mean
[04:20:52] <cradek> like a concave corner with an extra tiny move in it
[04:21:14] <cradek> I think the CAM always does the comp for that reason
[04:21:30] <cradek> and I think small-comp-from-nominal will work very reliably now
[04:22:35] <SWPadnos> with the assumption that it's always undersize
[04:22:45] <cradek> no, it can be either
[04:23:03] <SWPadnos> if you swap in a 1/4" mill for a 6mm one, you could still end up with gouging
[04:23:30] <SWPadnos> in the cases where the mill just fits through a gap that gets "closed" by a much later segment
[04:23:36] <cradek> sure
[04:23:39] <SWPadnos> (the case with arc 3 as jmk pointed out)
[04:23:49] <cradek> that's not what I mean though by "localized"
[04:24:01] <cradek> but I see your point
[04:24:04] <steve_stallings> Adding gouge handling to the comp code is a significant accomplishment, but like many things, people will only mention it when it doesn't do what they want. 8-)
[04:24:19] <cradek> a different kind of gouging (hitting some other remote part of the work)
[04:25:05] <cradek> steve_stallings: at work, my boss gives me a great review if he hasn't heard anyone complain about me recently - he understands that
[04:25:18] <cradek> since that's the best compliment possible
[04:25:27] <SWPadnos> smart boss
[04:26:06] <SWPadnos> o bad customers aren't like that ;)
[04:26:57] <SWPadnos> uh. too bad ...
[04:27:06] <SWPadnos> I wonder where my to went
[04:28:10] <steve_stallings> ver there
[04:28:11] <steve_stallings> o
[04:28:18] <SWPadnos> oh, thanks
[04:40:50] <steve_stallings> steve_stallings is now known as steves_logging
[08:52:34] <alex_joni> SWPadnos: you're not still around by any chance?
[09:46:21] <micges> hi all
[09:48:22] <alex_joni> hi
[09:51:04] <micges> I have one big reqest: could patch allowing translating rs274ngc error messages be merged from TRUNK to 2.2_branch ?
[09:52:41] <micges> cvs server is not working also
[10:00:58] <alex_joni> micges: hmm.. how extensive is the change for translating error messages?
[10:02:16] <micges> I saw that was few lines to correct working _ function
[10:02:49] <micges> wait
[10:03:07] <alex_joni> http://sourceforge.net/tracker/index.php?func=detail&aid=1650938&group_id=6744&atid=106744
[10:03:22] <alex_joni> beeing a bugfix, I think it could be considered
[10:04:08] <micges> 2.3 in a 4 month, merge will be very helpfull
[10:07:01] <alex_joni> the change is not that extensive (in emctaskmain()), although it changes the runscript .. probably also safe
[10:07:17] <alex_joni> micges: I'll talk to jepler about this (as he's the release manager for 2.2)
[10:08:02] <alex_joni> for reference emctaskmain.cc 1.121 .. 1.123
[10:08:15] <micges> I see
[10:08:21] <micges> thanks
[10:08:35] <alex_joni> configure.in 1.151 .. 1.152
[10:08:43] <alex_joni> config.h.in 1.17 .. 1.18
[12:54:16] <jepler> cvs service should now be restored.
[12:54:45] <micges> yes it is, thanks
[12:54:48] <jepler> bbl
[12:56:54] <alex_joni> jepler: ping when you're around
[13:40:59] <alex_joni> micges: did you test the translation of messages on TRUNK?
[14:06:13] <micges> nope
[14:06:53] <micges> I'm testing it now
[14:09:42] <alex_joni> I'll test later tonight (after 10pm I'm home.. so maybe then)
[14:10:01] <alex_joni> the most interesting part is if it works for installed versions, not just run-in-place
[14:11:20] <micges> after "make install" ?
[14:14:28] <alex_joni> micges: or after building a deb and testing that
[14:14:42] <micges> alex_joni: it's working in run-in-place here
[14:14:45] <alex_joni> make install has the slight disadvantage that you can't uninstall
[14:16:43] <alex_joni> you shouldn't use "make install" on a machine where you already have an installed emc2
[14:17:11] <micges> I know it will make mess :)
[15:03:44] <BigJohnT> http://www.cnczone.com/forums/showthread.php?p=549388#post549388
[15:10:45] <cradek> oh neat, there is now a /etc/apt/sources.list.d - he is right that we should use that
[15:11:39] <cradek> if he wants to share details about his project, our wiki is available
[15:11:53] <alex_joni> cradek: ping
[15:12:17] <BigJohnT> cradek: I'll mention that to him
[15:12:31] <cradek> and of course, usual disclaimer about preferring our lists to cnczone
[15:13:28] <BigJohnT> already did that :)
[15:17:05] <cradek> Personal wish list:
[15:17:05] <cradek> Rapid override (this exists on the formal feature request list)
[15:17:06] <cradek> The ability to select larger co-ordinate display using Axis.
[15:17:14] <cradek> someone's wish list is out of date...
[15:18:23] <micges> bbl
[15:42:59] <alex_joni> rofl "Benny offered his free time to create a forum , something that would have been a great benefit for those of us who would have liked an alternative to the primitive other options ."
[15:43:14] <alex_joni> (where primitive other options referrs to mailing list and IRC)
[15:43:44] <BigJohnT> lol
[15:45:50] <alex_joni> * alex_joni sighs
[15:45:53] <alex_joni> later all
[15:46:27] <BigJohnT> later
[15:57:19] <jepler> bah, benny. benny offered to create a forum where he would be a very important man and collect advertising revenue
[15:57:26] <jepler> all the while showing that he Didn't Get It
[16:21:20] <alex_joni> otoh, we already have a forum that's set up, but chose not to enable it
[16:23:28] <alex_joni> although it seems that a "vast" majority of 46% would prefer a forum : http://www.linuxcnc.org/component/option,com_poll/task,results/id,4/lang.en/
[16:23:41] <alex_joni> (I know 28 votes isn't actually useable data .. :)
[16:24:19] <BigJohnT> how many times did Benny vote
[16:24:34] <alex_joni> and over 2 years :D
[16:24:59] <alex_joni> maybe we should let the new Board discuss this again :D
[16:32:43] <alex_joni> BigJohnT: if you're not so busy later prod me
[16:32:52] <cradek> or not
[16:32:54] <alex_joni> I'll be back in about 3h or so
[16:33:14] <alex_joni> cradek: well.. maybe we're not part of it by then :)
[16:33:28] <alex_joni> bbl
[16:34:20] <cradek> quite possible
[16:34:53] <jepler> alex_joni: I'm sure there's a big correlation between users who want web forums and users who answer web polls
[16:38:05] <SWPadnos> maybe you could get an economic stimulus grant to do a study on that
[20:30:05] <CIA-1> EMC: 03cmorley 07TRUNK * 10emc2/src/hal/classicladder/ (5 files): Add support for connecting to HAL float pins (%If and %Qf) though ladder chops them as integers right now
[20:34:14] <CIA-1> EMC: 03cmorley 07TRUNK * 10emc2/src/hal/classicladder/module_hal.c: add support for floats. fix an error when registering hal s32 pins used 'bit' placeholder instead of 's32'
[20:40:03] <CIA-1> EMC: 03cmorley 07TRUNK * 10emc2/src/hal/classicladder/ (3 files): Add capabillity to change the databits,stopbits and parity of modbus serial port-This info is saved and loaded but cannot be changed in the GUI yetfiles.c
[20:40:47] <CIA-1> EMC: 03cmorley 07TRUNK * 10emc2/src/hal/classicladder/files.c: Add capabillity to change the databits,stopbits and parity of modbus serial port-This info is saved and loaded but cannot be changed in the GUI yet
[22:02:44] <cradek> dang, a bug disappeared when I recompiled from clean with -O0 (and then was still gone when I changed back to normal)
[22:03:00] <skunkworks_> yeck
[22:03:59] <alex_joni> nice way of getting rid of bugs
[22:04:06] <alex_joni> can you try for the rest of them?
[22:04:35] <alex_joni> whee it works for RIP
[22:04:40] <alex_joni> now to test installed
[22:05:06] <alex_joni> hi BigJohnT
[22:05:14] <cradek> http://timeguy.com/cradek-files/emc/lathe_concavecorners.png
[22:05:38] <BigJohnT> hi alex_joni
[22:05:52] <BigJohnT> what's up
[22:06:20] <alex_joni> BigJohnT: not much.. trying to get a nice patch for jepler to include in 2.2
[22:06:49] <BigJohnT> what kind of patch?
[22:07:03] <alex_joni> translations for rs274 error messages
[22:35:29] <alex_joni> yay, it works from emc2_2.2.9~cvs_i386.deb too
[22:35:33] <alex_joni> jepler: patch coming up
[22:36:56] <skunkworks_> cradek: what am I looking at?
[22:37:08] <alex_joni> concave corners on a lathe
[22:37:09] <alex_joni> :D
[22:37:39] <skunkworks_> huh
[22:38:24] <alex_joni> wicked, eh?
[22:38:28] <skunkworks_> neat
[22:38:35] <skunkworks_> lots of work :)
[22:39:47] <cradek> skunkworks_: programmed path (lots of sharp corners) and compensated path (curvy because the tool has a radius)
[22:40:12] <skunkworks_> very nice. :)
[22:40:13] <skunkworks_> bbl
[22:41:43] <CIA-1> EMC: 03cradek 07TRUNK * 10emc2/src/emc/rs274ngc/ (interp_convert.cc interp_queue.cc interp_queue.hh): detect localized gouges and try to explain it with an error message
[22:42:46] <CIA-1> EMC: 03alex_joni 07v2_2_branch * 10emc2/src/hal/drivers/mesa7i43-firmware/.cvsignore: more silencing
[22:45:40] <alex_joni> jepler: http://pastebin.ca/1304678
[22:51:40] <jepler> alex_joni: e-mail a copy please
[22:52:40] <alex_joni> ok, will do
[22:53:06] <alex_joni> attachment or inlined?
[22:53:21] <jepler> whichever will apply cleanly
[22:55:11] <jepler> that's up to you
[23:37:42] <CIA-1> EMC: 03cradek 07TRUNK * 10emc2/src/emc/rs274ngc/interp_queue.cc: oops, remove debug output
[23:58:34] <jepler> alex_joni: I got the patch from you