#emc-devel | Logs for 2007-08-25

[00:24:47] <jepler> cradek: well I've got a solution that seems to work for plain G43 offsets (in Z and X but I only tested Z)
[00:24:56] <jepler> the dimension still turns red but it runs without warning
[00:25:13] <cradek> yay!
[00:25:21] <jepler> since (I think) probes always go to the very bottom it might do something mostly useful in that case
[00:25:27] <jepler> let me check it in, as always you get to test
[00:25:35] <cradek> will do
[00:25:43] <jepler> it involves storing more information and doing more work at load time, so it's not a free lunch
[00:25:53] <cradek> slow but right is better than fast but wrong
[00:28:09] <jepler> committed
[00:28:19] <jepler> looks like cia is slow or dead
[00:28:20] <cradek> thank you! I'll try it.
[00:28:52] <cradek> if you mark the bug fixed, maybe petev will test it too
[00:33:39] <jepler> done
[00:40:37] <cradek> jepler: you checked in debug output in emccanon.cc
[00:43:03] <jepler> cradek: oh crap
[00:43:58] <jepler> fixing it as we speak
[01:12:23] <cradek> testing...
[01:20:58] <cradek> jepler: yay, I think you fixed both of the bugs that were getting in the way of my program
[01:21:31] <jepler> both?
[01:21:53] <cradek> yeah the other was the velocity mode stuff
[01:22:29] <jepler> I didn't fix any velocity mode stuff
[01:22:35] <jepler> you fixed that, didn't you?
[01:22:36] <cradek> yes you did, a week or so ago
[01:22:53] <cradek> rapids/tool change moves
[01:22:56] <jepler> oh -- sorry, I was thinking of the jog wheel thing you fixed
[01:25:40] <cradek> this is a neat little program - you should see it sometime
[01:26:29] <jmkasunich> cradek: whats the name of that surplus place by you?
[01:26:32] <cradek> surplus center
[01:26:51] <jmkasunich> thanks
[01:27:11] <jepler> cradek: what is the program?
[01:27:33] <jepler> I am looking at http://sourceforge.net/tracker/index.php?func=detail&aid=1734299&group_id=6744&atid=106744 and I can't see a bug
[01:28:37] <jepler> on the surface petev's claim that "emc should never command a position outside of [soft] limits" sounds like a good one, but the alternative is to command a step to just inside the limit, which will not do something you want
[01:29:04] <jepler> if he means the machine should not turn on, I think he's mistaken because the way to fix a machine on a soft limit is to simply jog it back
[01:30:16] <jmkasunich> you should not set a home position that is balanced on the razor's edge of falling outside the limit
[01:30:24] <jmkasunich> pete is being difficult
[01:30:49] <cradek> fwiw, my BP is also that way
[01:30:54] <jepler> what way?
[01:30:56] <cradek> it homes to 0.000 which is also the lower limit
[01:31:09] <jmkasunich> bah
[01:31:36] <cradek> (on the quill it homes to 0.000 which is also the upper limit)
[01:33:31] <jepler> after homing it sits right at 0.000 ?
[01:33:33] <cradek> well now that I think about it, I don't know how it does it, since it never backs up
[01:33:38] <cradek> jepler: yes
[01:34:00] <cradek> I guess the home position must be past the index by some fixed amount
[01:44:20] <jepler> one of you guys should try closing it again -- I did and it failed to stay closed
[01:45:17] <cradek> not me, since I'm not sure it's wrong
[01:49:46] <cradek> well I can get it too on my lathe. it's easy enough to work around but I think I agree with pete that I don't see the problem with this configuration
[01:51:25] <cradek> hmm, it does freak out after that too
[01:51:36] <cradek> jogs that should be continuous are only moving .001 mm
[01:51:58] <cradek> and, if I poke Home again, the "homed" icon goes away but it doesn't home again
[01:53:52] <cradek> if I turn machine off then on again, it goes back to normal
[01:54:51] <cradek> if I set the max limit to 40.50001 (HOME is 40.5) it works like I expect
[01:55:39] <SWPadnos> maybe just add/subtract $TINY_NUMBER from the ini limits
[01:56:33] <cradek> hmm, cannot pick up chips 1' long with a shop vac
[01:56:42] <SWPadnos> either something that won't get lost in the double mantissa, or some multiple of 1/INPUT_SCALE
[01:56:44] <SWPadnos> heh
[01:56:48] <SWPadnos> needa bigger shop vac ;)
[01:57:11] <cradek> anyone need some aluminum wool?
[01:57:45] <SWPadnos> hmm. not at the moment
[02:14:07] <jmkasunich> wool? that comes from lathe, not mill
[02:37:47] <cradek> yep
[03:35:17] <cradek> hmm, I didn't do anything about teleop uvw
[03:56:38] <Skullworks-PGAB> - - snip - -
[03:56:40] <Skullworks-PGAB> But if you want good code, look to emc, its the best tutorial in the world on how to write a cnc
[03:56:42] <Skullworks-PGAB> program, but will take you a few weeks to truly understand how it functions in all its various areas.
[03:56:42] <Skullworks-PGAB> The EMC programmers are the most dedicated and selfless group you will find in the cnc world.
[03:56:42] <Skullworks-PGAB> Reading their code, and understanding their workflow is the first , best step to understanding how
[03:56:42] <Skullworks-PGAB> all cnc machines work. Read my manuals as well, they try to gently ease you into it, and are used all
[03:56:42] <Skullworks-PGAB> over whether with or without mach3, simply as a starting users guide to cnc stuff.
[03:57:21] <Skullworks-PGAB> If your intent is to send to something like a G100 or serial device, the job is much easier, and the EMC code,
[03:57:21] <Skullworks-PGAB> being open source, is a true university of cnc programming. I think I have my doctorate in EMC, and it was
[03:57:21] <Skullworks-PGAB> a fantastic learning experience..
[03:57:21] <Skullworks-PGAB> Thanks,
[03:57:22] <Skullworks-PGAB> Art
[03:57:25] <Skullworks-PGAB> www.artofcnc. ca
[03:58:20] <Skullworks-PGAB> Art (mach 3) recommending EMC sorce as the best training tool.
[04:06:16] <LawrenceG> cool... nicely said
[11:47:51] <fenn> shouldnt there be a motion.analog-in-NN for m66 E?
[11:56:16] <fenn> ah "* analog part still missing"
[12:42:10] <fenn> axis.N.joint-pos-fb is an OUT pin; shouldnt it be an IN pin?
[12:44:26] <jepler> no
[12:44:31] <jepler> joint-pos-fb is motor-pos-fb with the joint offset subtracted out
[12:44:58] <fenn> what is joint-pos-fb used for?
[12:45:00] <jepler> suppose you want to do "something" based on the joint position .. you use that pin as your input
[12:46:16] <fenn> joint offset is like, backlash comp and screw error comp right?
[12:46:27] <jepler> and the offset created by homing
[12:49:08] <jepler> oh I forgot the main reason but I rediscovered it when I read the CVS log
[12:49:09] <jepler> convert joint_pos_fb and joint_pos_cmd from parameters to pins so that these
[12:49:11] <jepler> values can be used for machine visualization
[12:49:26] <fenn> ah
[12:49:36] <jepler> when you are running the puma visualization program you need joint positions, not motor positions