#emc-devel | Logs for 2009-05-15

[12:26:17] <SWPadnos> micges, thanks
[12:35:18] <micges> sure
[13:17:50] <micges> SWPadnos: have used mesa stepgens?
[13:17:57] <SWPadnos> nope
[13:18:20] <micges> I found many bugs in it today :|
[13:18:25] <SWPadnos> hmmm
[13:18:41] <micges> machine is unusable, must switch to s/w stepgens
[13:18:46] <SWPadnos> go ahead and describe them. someone will notice
[13:19:10] <micges> I have team on place of error, they gathering information
[13:19:15] <SWPadnos> do you have the latest stepgen firmware?
[13:19:39] <SWPadnos> if these machines are using packaged EMC2, the firmware is a separate package for 2.3
[13:19:51] <micges> all my machines are RIP
[13:20:08] <micges> cvs 2.3 branch from about 5.05
[13:20:13] <SWPadnos> ok, did you set up a link in /lib/firmware to point to your RIP firmware?
[13:20:21] <SWPadnos> or copy the appropriate files
[13:21:04] <micges> link
[13:21:47] <SWPadnos> hmm
[13:21:47] <micges> haw can I be sure about version of firmware ?
[13:21:58] <micges> any status command ?
[13:22:05] <SWPadnos> I don't think so
[13:22:18] <SWPadnos> dmesg should show the name of the file that gets loaded
[13:28:40] <micges> bitfile SVST8_4.BIT is last commited 4 months ago
[13:29:03] <micges> so it is in firmware debs and is cvs 2.3.0
[13:29:27] <micges> so with and without linking these bitfiles are the same
[13:29:56] <jepler> I must have missed it -- what is the actual problem?
[13:30:36] <micges> while jogging >6000 mm/min mesa stepgens reverse direction randomly
[13:30:48] <micges> measured on electric cables
[13:31:08] <micges> there are two axes
[13:31:40] <micges> the same configuratuion (updated after last report about negative scales)
[13:31:45] <micges> X works fine
[13:31:56] <micges> on Y there are above issues
[13:32:22] <micges> next behaviour: loosing steps heavly
[13:32:31] <micges> many mm are lost
[13:33:19] <micges> badly I don't have any machine to test :|
[13:34:08] <micges> and after reversion of dir signal it stays inverted
[13:34:18] <micges> jogging directions are switched
[13:34:28] <cradek> !!
[13:34:36] <micges> until restart of emc
[13:35:35] <cradek> does it happen reliably? can you say how to make it happen?
[13:36:25] <micges> in details I can tell tomorow or so
[13:36:46] <jepler> I recommend filing it as a bug on sourceforge
[13:37:05] <jepler> the hostmot2 maintainer isn't here, and the sourceforge tracker will let you attach files if necessary
[13:37:05] <micges> now I can tell that on jogging at high speed above whole table on some places dir reverts
[13:37:24] <micges> I know
[13:37:42] <micges> when I'll have all information I'll fill bug
[13:48:37] <BJT-Work> micges: are you still around?
[13:48:44] <micges> yes
[13:49:18] <BJT-Work> I'm fixing to use a mesa 5i20 on my plasma cutter to drive my gecko step drives.
[13:49:47] <BJT-Work> just reading back your running steppers and see the problem
[13:50:21] <micges> want config to see?
[13:50:27] <BJT-Work> yes
[13:50:39] <micges> wfm
[13:51:53] <micges> ini: http://www.pastebin.ca/1423678
[13:52:45] <micges> hal: http://www.pastebin.ca/1423680
[13:52:57] <micges> only X Y are used
[13:53:02] <BJT-Work> ok
[13:53:04] <micges> rest you can ingore
[13:53:57] <BJT-Work> does the DRO show the proper travel when you missed steps?
[13:54:15] <micges> yes
[13:54:41] <BJT-Work> ok
[13:55:41] <jepler> so to reproduce your problem, use a real scope on the "x direction" physical output, jog towards +ve at 6000mm/min, and observe that the direction bit flips on the scope?
[13:56:23] <micges> on y direction, yes
[13:56:45] <micges> problem is on Y axis
[13:56:58] <BJT-Work> what is your scale for each axis
[13:57:08] <BJT-Work> nm
[13:57:33] <micges> I'm going home
[13:57:37] <micges> will be later
[13:57:41] <BJT-Work> ok
[13:57:42] <micges> see you
[13:57:53] <jepler> the last time we were talking, I thought the problem was concerning a negative axis SCALE. I don't see a negative SCALE in your inifile. Did you change the motor wiring to get rid of the need for the negative scale?
[13:57:57] <jepler> argh
[13:59:23] <jepler> #
[13:59:26] <jepler> setp hm2_5i20.0.gpio.055.invert_output 1
[13:59:26] <jepler> 15 55 IOPort StepGen 1 Dir/Table2 (Out)
[13:59:32] <jepler> I see he did something else
[14:01:40] <jepler> I see his threads are still configured wrong: hm2_5i20.0.read should probably be the very first thing, but it's after the motion controller
[14:02:44] <jepler> but those shouldn't cause this problem..
[14:07:58] <BJT-Work> does the order of adding a function relate to the order of execution?
[14:08:15] <SWPadnos> yes, unless you specify a position in the addf line
[14:08:23] <BJT-Work> ok
[14:52:56] <alex_joni> logger_dev: bookmark
[14:52:56] <alex_joni> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2009-05-15.txt
[14:52:59] <alex_joni> micges: ^
[14:54:10] <micges> jepler: config was changed that negative scale is not used
[14:54:28] <micges> I don't know exact change becouse not I changed it
[14:57:13] <micges> jepler: I know threads are configured wrong but those shouldn't cause this problem
[14:57:41] <micges> I didn't have chance to correct threads in that machine
[14:57:51] <jepler> revision control software like git, bzr, or rcs are easy to set up (no server required, and a whole-directory copy copies history information too) and can track changes to configuration files like emc
[14:58:05] <jepler> I heartily recommend their use to keep track of configuration changes, particularly if more than one person is making them
[15:08:39] <micges> jepler: I have subversion to control interface source
[15:09:31] <micges> not configs :|
[15:09:53] <SWPadnos> some people recommend using cvs (or similar) for /etc :)
[15:09:54] <jepler> you'd probably benefit from putting configurations in svn too
[15:11:17] <micges> ok question: how would you maintain config of machine that was selled ?
[15:11:31] <micges> repo on flash disk ?
[15:13:06] <micges> SWPadnos: cool idea ;)
[15:16:15] <jepler> with decentralized versioning systems it could work
[15:16:48] <micges> ugh I must read more of that :)
[15:16:50] <jepler> the service tech carries a copy of the repository to the user site and can use it to apply updates or record changes
[15:16:59] <jepler> of course, he'll have to know a lot about git
[15:17:25] <jepler> bbl
[16:33:00] <micges> for translation: in halscope, what mean AC coupling option in offset window in mode settings ?
[16:33:26] <micges> in vertical settings ?
[16:34:14] <cradek> it means center the trace on the zero line, no matter what the real value
[16:34:27] <cradek> like AC coupling on a real scope
[16:35:50] <micges> ah ok
[19:06:59] <CIA-67> EMC: 03seb 07TRUNK * 10emc2/src/hal/drivers/mesa-hostmot2/stepgen.c: commit "stepgen oscillation with negative scale" fix from Jeff Epler
[19:11:54] <CIA-67> EMC: 03seb 07v2_3_branch * 10emc2/src/hal/drivers/mesa-hostmot2/stepgen.c: merge Jeff's "stepgen oscillation with negative scale" fix from Trunk
[19:13:08] <seb_kuzminsky> jepler: i tested your hm2 stepgen patch and it looks like it fixed the problem
[19:13:15] <seb_kuzminsky> i've committed it to Trunk and 2.3
[19:15:55] <CIA-67> EMC: 03seb 07v2_3_branch * 10emc2/debian/changelog: note hm2 stepgen fix
[19:16:57] <jepler> seb_kuzminsky: great, thank you for testing it
[19:17:09] <seb_kuzminsky> no, thank *you* for fixing it ;-)
[19:17:18] <jepler> seb_kuzminsky: did you see micges talking about some other possible problem earlier today?
[19:17:27] <jepler> it sounds like we don't have much information yet
[19:17:41] <seb_kuzminsky> logger_dev: bookmark
[19:17:41] <seb_kuzminsky> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2009-05-15.txt
[19:18:01] <jepler> 08:29:56 <jepler> I must have missed it -- what is the actual problem?
[19:18:02] <jepler> 08:30:36 <micges> while jogging >6000 mm/min mesa stepgens reverse direction randomly
[19:18:05] <jepler> 08:30:48 <micges> measured on electric cables
[19:18:37] <jepler> he seemd to have switched from negatve scale to inverted direction output according to the hal files he posted; I dunno whether that is relevant or a green herring
[19:19:17] <jepler> so-o -- can you decide today whether you want me to hold 2.3.1 for this bug?
[19:21:06] <micges> on monday I will have brain storm with my service team
[19:21:48] <seb_kuzminsky> micges: can you test the current Trunk or 2.3 branch? I just committed Jeff's patch that i think fixes the oscillation that you and Matt Shaver reported
[19:21:58] <micges> I will gather all information and/or test it on mesa with motors but not on real machine
[19:22:23] <seb_kuzminsky> you dont even motors, you can look at halscope to see if the motion is correct
[19:23:19] <micges> I have friday evening now here :|
[19:23:19] <seb_kuzminsky> micges: let me see if i understand the problem your're reporting
[19:23:34] <seb_kuzminsky> oh, time for something other than debugging machines ;-)
[19:23:51] <seb_kuzminsky> if you jog X, everything is fine
[19:24:25] <seb_kuzminsky> if you jog Y at 6000 mm/min, it randomly reverses direction, then doesnt "unreverse" direction to the correct value until you reboot the machine?
[19:24:40] <micges> yes
[19:25:16] <jepler> do you know for sure whether any of these are important: jog vs g0? 6000mm/min vs 5000mm/min? "invert" of direction pin using gpio invert?
[19:26:02] <micges> jepler: question for me ?
[19:26:07] <jepler> yes
[19:28:39] <micges> I know that switching to software stegens remove all issues immadietly
[19:28:48] <seb_kuzminsky> it's suspicious to me that a scope on the Y Dir line coming out of the 5i20 shows the dir flip
[19:28:57] <seb_kuzminsky> micges: can you step at 6000 mm/min with sw stepgen?
[19:29:30] <micges> nope
[19:29:53] <micges> about 4000
[19:30:01] <seb_kuzminsky> what happens if you command the hm2 stepgen to step and the highest rate that you've verified with sw stepgen?
[19:30:49] <micges> I cannot test it
[19:30:55] <micges> bbl
[19:31:04] <seb_kuzminsky> ok, have a good friday night micges :-)
[19:31:12] <seb_kuzminsky> we'll get this fixed
[19:31:30] <jepler> first step is to unmask the real problem
[19:31:37] <jepler> find out what is essential to cause it
[19:32:16] <seb_kuzminsky> i will try micges' config on my test system soonish and see if i can reproduce it
[19:32:29] <seb_kuzminsky> i think you shouldnt hold up 2.3.1 for this
[19:35:09] <jepler> OK
[19:35:20] <jepler> if you find you're hot on the trail of it saturday, drop in and let me know
[19:35:37] <jepler> otherwise I'll probably tag and release saturday afternoon or sunday am
[19:36:33] <jepler> alex_joni: I can build all the dapper packages, but can you build the breezy packages once again?
[19:36:44] <jepler> er, er, I'm two years off
[19:36:52] <jepler> I can build all the hardy packages, but can you build the dapper packages once again?
[19:37:16] <alex_joni> sure
[19:37:23] <alex_joni> on sunday?
[19:37:28] <alex_joni> * alex_joni reads back
[19:37:34] <jepler> yeah probably
[19:37:42] <jepler> if it's a day late, who cares? not this guy!
[19:38:12] <alex_joni> I'll be away on saturday night, not sure how late I will be back
[19:38:26] <alex_joni> maybe I'll get them done on sunday morning
[19:39:19] <jepler> bbl
[19:40:41] <alex_joni> if any of you cares for a bbq and a beer, feel free to drop by on sunday
[19:41:24] <jepler> sounds nice, but I can't make it
[19:41:28] <jepler> software release has to be made
[19:43:36] <alex_joni> you can do it here ;)
[19:43:59] <alex_joni> there will be veggie satay
[19:44:24] <jepler> anyway, I've already been invited to Ingrid's folks' house for sunday dinner
[21:20:38] <micges> logger_dev:bookmark
[21:20:38] <micges> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2009-05-15.txt
[21:22:04] <micges> feature request:
[21:26:13] <micges> some special mode when Fnn doesn't work, and velocity commanded of emc is calculated from external hal pin of value let say 0..1
[21:27:02] <micges> so any user module can write to this pin in realtime and then can apply any sort of correction of velocity due to direction (for lasers)
[21:27:44] <micges> 0..1 <=> 0..[TRAJ]VEL_MAX
[21:28:38] <micges> I know that simmilar thing is in halui but I'm asking for feed rate veloctiy not machine velocity
[21:29:58] <micges> that allow also to change velocity wothout stopping/pausing/whatever program
[21:33:48] <micges> something simmilar to feedoverride in realtime
[21:34:11] <micges> when this mode is enabled, normal feedoverride should be disabled
[21:47:06] <jepler> whee!
[21:47:21] <jepler> * jepler runs hm2-stepper and looks at the blinkenlights on the 7i31 led card
[21:48:11] <BigJohnT> * BigJohnT likes blinky lights on his 7i31
[21:49:16] <BigJohnT> do you have a 5i20?
[21:49:25] <jepler> yeah this one is 5i20
[21:49:41] <jepler> got it 2+ years ago in the group buy; I've never used it for anything
[21:50:16] <jepler> micges: you gave us hal files that don't work without components we don't have
[21:50:20] <jepler> micges.hal:34: Can't find module 'vctrl' in /usr/realtime-2.6.24-16-rtai/modules/emc2
[21:50:39] <BigJohnT> I'm working on putting mine in my plasma cutter to drive the steppers and control the torch height
[21:51:09] <jepler> eventually I'm going to replace pluto with mesa for my circuit board milling machine..
[21:52:18] <BigJohnT> * BigJohnT wanders off to clean the battery box for the truck :(
[21:56:38] <jepler> well I got micges config to run, but after jogging for 5 minutes I didn't see anything weird in the LEDs
[21:57:13] <jepler> the "Y direction" LED is lit while stopped or travelling towards +Y; it's dark whe ntravelling towards -Y.
[21:57:20] <jepler> I never saw it reverse this relationship
[21:58:41] <BigJohnT> the only thing I noticed was the Y had a larger scale than the X...
[22:04:05] <jepler> yeah -- it shouldn't matter, though
[22:04:41] <micges> jepler: ignore vctrl
[22:05:00] <jepler> by the way I was testing with 2.3.0 package
[22:05:13] <jepler> bbl
[22:07:48] <CIA-67> EMC: 03jepler 07TRUNK * 10emc2/src/hal/utils/ (scope_horiz.c scope_vert.c): make additional strings translatable (thanks micges)
[22:09:41] <micges> jepler: bad you can't reproduce error