#emc | Logs for 2008-08-21

[00:58:21] <SWPadnos> jepler, I think someone mentioned that the tool table size had to do with the maximum number of some struct you could fit into some arbitrary amount of memory (like a 4k page or something)
[01:20:55] <owad> Can I compensate for backlash in emc2? I can't find a setting for it in stepconf wizard, or instructions in the manual.
[01:23:02] <SkinnYPup> there's a backlash setting in stepper_inch<or mm>.ini
[01:23:07] <SWPadnos> you can, but it's always recommended to fix it in hardware
[01:23:13] <SWPadnos> (ie, mechanically)
[01:26:14] <SWPadnos> incidentally, backlash is convered in the basic configuration page
[01:26:53] <SWPadnos> http://www.linuxcnc.org/docview/html//config_ini_config.html#sub:%5BAXIS%5D-Section
[01:30:37] <owad> ok, I found where the setting goes (the wizard stores it in a filed named [MACHINE_NAME].ini) and understand how to add the setting. Thanks, SkinnYPup and SWPadnos!
[02:13:03] <jepler> hah, the patch I submitted to pyserial was for a feature already implemented months ago. :) too bad that version didn't make it into hardy :(
[02:27:08] <cradek> jepler: funny
[02:28:17] <cradek> I've done that too - I fixed a bug in xscreensaver and even sent it to jwz - it had been fixed like a year ago
[02:28:49] <cradek> btw, I have X feedback! the resolver setup is working.
[02:30:49] <cradek> will an everyday cheapie voltmeter in AC mode show anything like rms for a 10kHz @ a-few-volts signal?
[02:31:53] <cradek> I mean a sine wave
[02:45:08] <jepler> cradek: oh cool
[02:45:48] <cradek> but no motion tonight. I know not to hook up finicky power supply stuff after 9pm.
[03:04:16] <CanSir> Can anyone suggest why I can see the inputs on an OR2 component go high, but the output always stays low?
[03:04:28] <cradek> you forgot to add its function to a thread
[03:04:41] <CanSir> Yes I did. Thanks!
[03:04:49] <cradek> I've done it many times :-)
[03:04:49] <CanSir> :redfaced:
[03:06:28] <CanSir> I'm using buttons on a joypad to control jog speed - which thread would you recommend I add it to? servo-thread?
[03:07:10] <cradek> halui will read the inputs much more slowly than the servo thread, so it seems like that should be fine
[03:07:41] <CanSir> Great - thanks for the help!
[03:07:58] <cradek> welcome
[03:16:31] <pminmo_> is there a recommended set of development tools for python, gtk, c......
[03:17:50] <LawrenceG> Jymm, JymmmEMC http://imagebin.ca/view/ugq67j2.html ttt and bezel code at work
[03:18:28] <cradek> neat!
[03:18:37] <cradek> making a one tube regenerative?
[03:19:08] <LawrenceG> yea... the old fashioned way... on a real piece of oak
[03:19:16] <cradek> I have a couple nice old variometers (is that the right word?) and I will do that one of these days
[03:19:23] <LawrenceG> front panel will be Al
[03:19:59] <LawrenceG> yup.... I did a bunch of work on ttt to support runtime scaling, rotation and translation
[03:20:23] <cradek> Al will make a nice hand shield
[03:20:47] <LawrenceG> do you have email... I think its a good mod for ttt
[03:20:55] <cradek> chris@timeguy.com
[03:20:59] <LawrenceG> I'll send you the file
[03:21:00] <cradek> thanks, I will look it over
[03:21:17] <LawrenceG> or I could pastebin it
[03:21:28] <cradek> a patch in email is best
[03:22:34] <LawrenceG> not sure how well a patch will work.... I built on my existing version which was from before ttt v3
[03:22:47] <LawrenceG> the file is not that big...
[03:23:26] <cradek> that's ok, I'll see what I can do.
[03:24:44] <LawrenceG> 14k on its way
[03:25:15] <cradek> thanks
[03:25:34] <LawrenceG> did you see http://imagebin.ca/view/O3CsXvpg.html some of the scaling etc
[03:26:03] <LawrenceG> independant x and y scales
[03:26:27] <cradek> heh, you did the rotation in gcode
[03:26:44] <cradek> but you're right, your version looks nothing like the current version
[03:27:33] <LawrenceG> :} one gcode file for a specfic string..... rest is adjustable at runtime using vars
[03:28:45] <LawrenceG> subroutines rule!
[03:32:03] <LawrenceG> for the panel, I did up each the strings with ttt, pasted them into a gcode file and then moved them around by changing the vars using axis as a layout viiewer
[03:32:14] <jmkasunich> last two evenings of work: http://jmkasunich.com/pics/boring-insert-holder-2572.jpg
[03:32:31] <jmkasunich> uses SNMP-32x inserts
[03:32:43] <cradek> wow
[03:32:58] <LawrenceG> cool.... I love show and tell
[03:33:14] <cradek> does the insert have relief on the front so you can bore up to a square shoulder?
[03:33:19] <jmkasunich> no
[03:33:21] <cradek> or is it square?
[03:33:26] <jmkasunich> its a square
[03:33:28] <cradek> ah
[03:33:42] <cradek> I thought it was one of those barely-a-diamond ones
[03:33:43] <jmkasunich> these are inserts I already had
[03:33:58] <jmkasunich> I bought the 80 degree ones on ebay, they're not here yet
[03:34:03] <cradek> 3/4?
[03:34:07] <jmkasunich> they're also quite a bit smaller
[03:34:12] <jmkasunich> the bar in the pic is 3/4
[03:34:29] <jmkasunich> the insert in the pic is 3/8" square
[03:34:35] <jmkasunich> the ones I bought are 1/4"
[03:34:51] <jmkasunich> I have a bar for them, but its a lathe style bar - 6" long (carbide)
[03:35:06] <jmkasunich> this bar is for use in a boring head - it is just long enough for the job at hand
[03:35:25] <cradek> 3/4 is a big boring head
[03:36:14] <jmkasunich> yeah, I was considering using my dad's boring head: http://jmkasunich.com/pics/boring-head-2563.jpg
[03:36:16] <cradek> that's neat. I like it. seems like it would be hard to get the tip of the business end on the center line.
[03:36:40] <jmkasunich> but dmess talked me out of it (2.7" hole in cast iron)
[03:36:56] <jmkasunich> so I borrowed the old, ugly, but beefy one that we never use at work
[03:37:26] <jmkasunich> I did a layout in easy cad to get the tip height right
[03:37:27] <cradek> that one you have is neat, but not beefy
[03:38:03] <jmkasunich> the one I borrowed has a 3/4" shank and uses 3/4" tools
[03:38:20] <cradek> perfect
[03:38:21] <jmkasunich> and I found that I can grab it by the body in the 4-jaw, so the shank won't be the flimsy spot
[03:38:59] <jmkasunich> I'm going to be boring this part using the lathe spindle to hold the cutter and feeding the part toward it on the table
[03:39:48] <cradek> will your lathe go down to a few hundred rpm?
[03:39:54] <jmkasunich> 120
[03:39:58] <cradek> good
[03:40:10] <jmkasunich> I was pulling my hair out trying to figure out how to machine an insert pocket
[03:40:25] <jmkasunich> then I got inspired - make a flat for the insert, and put the pocket around it
[03:40:41] <jmkasunich> that is 0.062 O-1 ground flat stock locating the insert
[03:41:03] <cradek> that is clever
[03:41:16] <Guest147> hi
[03:41:20] <jmkasunich> hello
[03:42:51] <jmkasunich> am I that scary?
[03:43:36] <Guest147> sorry. looking up info on emc backlash comp if it exists. heard it was disabled in current release
[03:44:03] <jmkasunich> no, it is there
[03:44:28] <jmkasunich> it is disabled by default, but all you need to do is set a number in the ini file to tell EMC how much backlash you have, and it will attempt to compensate
[03:45:15] <Guest147> nice. I will look into that. I assume that stepconfig is useless once you edit the file directly?
[03:45:34] <jmkasunich> I suspect so
[03:46:00] <jmkasunich> I don't know if stepconfig can set the backlash number, or if that requires manual editing
[03:46:26] <jmkasunich> I'm old-fashioned and do things by hand, so I'm afraid I don't know stepconf very well - it may be able to do more than I'm aware of
[03:46:33] <jmkasunich> it is steadily improving
[03:46:41] <Guest147> thats fine. time to loose the training wheels. I think it needs a manual edit.
[03:47:23] <Guest147> I allways figured stepconf sets the basics, then manual to get to the nitty gritty
[03:47:52] <LawrenceG> off to make chips :}
[03:48:11] <jmkasunich> happy chipping
[03:48:15] <Guest147> it sure did. thanks
[03:48:35] <jmkasunich> I'm off to walk the dog and then get some sleep, goodnight all
[03:48:41] <Guest147> 'nite
[03:50:49] <Guest147> Guest147 is now known as topls
[03:54:41] <topls> anyone use the cncfusion kit (regular) for X3 mill?
[04:06:43] <toastyde1th> ffffff
[04:15:43] <topls> uuuuuuuu ?
[04:16:09] <topls> sorry. had to.
[05:26:03] <CanSir> Is there a way to have EMC just pause the run (no modal msg popup window) when it encounters a Tx command?
[08:20:15] <pjm> good morning
[08:21:39] <archivist_ub> good moaning
[08:24:56] <pjm> hehh yeah i feel like that! i'm back at 'work' today after 2.5 weeks vacation
[08:25:04] <pjm> needed to use holiday up or loose it
[08:29:41] <archivist_ub> hmm better think about doing a small amount of work, but not too much to stress myself
[08:32:57] <pjm> hehh i'm easing back into work by doing expenses!
[09:48:18] <archivist_ub> pjm Flight Refuelling is selling some cnc gear viewing Monday 15th sep
[09:50:02] <pjm> really
[09:50:06] <pjm> that's just up the road!!
[09:50:21] <pjm> if its the cobham road site, its less than 1 mile away from here
[09:50:36] <pjm> my landrover came from FR too!
[09:50:47] <archivist_ub> yup its in machinery classified thats just come
[09:51:37] <archivist_ub> www.apexauctions,com
[09:52:01] <pjm> cool thanks for that
[09:52:12] <pjm> i'm on holiday that week too!
[09:52:39] <archivist_ub> another holiday!!!!!
[09:55:42] <pjm> yeah i have to use it all up
[09:55:45] <pjm> or loose it
[09:56:07] <pjm> i did ask to take monday and friday off for a few months
[09:56:13] <pjm> but they said that would be impractical
[09:59:45] <pjm> archivist_ub btw there is a possibility that i will have a day out to warwick next week
[09:59:56] <archivist_ub> ok
[10:00:17] <archivist_ub> should be easy for me to escape
[10:05:43] <pjm> any particular day better?
[10:05:51] <pjm> except mon/fri
[10:07:30] <archivist_ub> steam engine driving monday
[10:07:40] <archivist_ub> any other day
[10:08:02] <archivist_ub> boss will get TOLD when im off :)
[10:08:25] <archivist_ub> he is not in my good books at the moment
[10:10:08] <pjm> hehh
[10:11:05] <archivist_ub> and I have now found out how much he is selling the site for, a paltry 130k
[10:11:28] <pjm> my missus reckons i need to get a industrial unit
[10:11:39] <pjm> the problem is i'd fill it with machines
[10:11:47] <archivist_ub> hehe so would I
[10:12:28] <archivist_ub> I need enough height for the toy lorry in my unit
[10:13:19] <Vq^> is it an error to use more than one code from the non-modal group? (group0, g4,g10,etc)
[10:42:42] <micges> good morning all
[10:43:05] <Vq^> morning mr micges
[10:44:54] <micges> any ideas how to fix bug sf#2046178
[10:44:56] <micges> ?
[15:43:09] <cradek> has anyone used cnc4pc's charge pump board?
[15:43:54] <cradek> the "docs" say more about mach than the product itself
[15:44:47] <cradek> "The parallel port can do weird things when the program is not in control. It could start your spindle, open you [sic] collect [sic], or make loose [sic] steps."
[15:45:24] <SWPadnos> heh
[15:45:37] <skunkworks> wait - did I write that?
[15:45:45] <cradek> "Stops your system if windows [sic] crash. [sic]"
[15:45:51] <SWPadnos> I don't know anyone who has used it with EMC, but there are several people on the geckodrive and CCED Yahoo groups who have
[15:46:08] <SWPadnos> "Stops your system if [sic] windows [sic] crash. [sic]" :)
[15:46:13] <cradek> haha
[15:47:10] <cradek> it doesn't say what the delays are, or what frequencies turn it on or off, or what the ratings on the relay are. one part of the manual says it runs on 5v and another part says 12v.
[15:50:32] <BigJohnT> cradek: ask him to send you a demo and we will document it
[15:50:53] <SWPadnos> that's not worth the $22.50 IMO
[15:50:59] <cradek> also I'm trying to pick between two MPGs. They are called "Rotary manual pulse generator" and "Larger size rotary manual pulse generator". guess what information I want to help me choose which one I want?
[15:51:03] <BigJohnT> he sent me a pendant to test... I hope Ray gave it back to him
[15:51:22] <SWPadnos> jmk had a schematic of the CP he made - maybe dig around for that one
[15:52:21] <BigJohnT> hmmm what the diameter is? do I get the prize?
[15:52:49] <cradek> if I had docs and knew it would do what I want, the price is perfectly fine I think.
[15:53:21] <cradek> as a throw-in when ordering some wheels, which is what I really want
[15:53:31] <SWPadnos> he has one on eBay for $10 less
[15:53:40] <SWPadnos> I just noticed it last night
[15:53:42] <cradek> yeah I saw that, I want two
[15:53:57] <SWPadnos> heh - buy the one, and I'll bet there's another one on eBay the next day
[15:54:12] <SWPadnos> there was someone selling two AB wheels for $119 I think
[15:54:13] <cradek> and pay for two shippings...
[15:54:19] <SWPadnos> (used of course)
[15:54:26] <cradek> oh? hmmmmm
[15:55:40] <jepler> cradek: if only you could read .dwg file, those two items have a link called "dimensions" ..
[15:56:11] <jepler> s/file/files/
[15:56:43] <SWPadnos> http://cgi.ebay.com/Lot-of-2-FANUC-MPG-Pulse-Generator-A860-0200-T020_W0QQitemZ130220283385QQcmdZViewItem?_trksid=p3286.m20.l1116
[15:58:42] <cradek> interesting, thanks
[15:59:29] <SWPadnos> sure
[16:05:22] <jepler> cradek: if you end up ordering new and no more expensive than the cnc4pc ones, order an mpg for me too.
[16:05:42] <jepler> (well, or used if you find a lot with more than you need)
[16:06:31] <cradek> ok
[16:30:06] <anonimasu> hello
[16:44:38] <cradek> now there is a nice fix. http://i16.ebayimg.com/06/i/001/03/87/47ad_1.JPG
[17:07:23] <anonimasu> heh..
[17:07:25] <anonimasu> death.
[17:07:33] <anonimasu> when it detaches at 10krpm.
[17:07:36] <anonimasu> ZOOM
[17:41:57] <OoBIGeye> i would be more worried about the machine walking out of the shop if that got up to 10k rpm...
[17:42:03] <OoBIGeye> un balanced anyone?
[17:54:22] <Lerman> Anyone know where I can find documentation on usrmot?
[17:55:17] <SWPadnos> have you tried the obvious: `man usrmot` ?
[17:55:28] <SWPadnos> (I don't know that it will work - just asking)
[17:55:48] <Lerman> Yes. Also searched user and integrator's manuals.
[17:56:54] <Lerman> From the integrator manual: The EMCMOT utility program USRMOT can be used to set these interactively and verify their results so that the proper values can be put in the INI file with a minimum of trouble.
[17:57:41] <SWPadnos> hmmm
[18:00:35] <SWPadnos> well, I'm not sure where the docs are, but you can run it (with EMC already running) with "usrmot -ini /path/to/ini"
[18:00:41] <SWPadnos> typeing ? will give a list of commands
[18:01:00] <SWPadnos> (it looks like a command shell)
[18:01:41] <Lerman> Thank you. That was obvious (not).
[18:01:48] <SWPadnos> heh
[18:02:02] <Lerman> I tried usrmot -help. That was useless.
[18:02:05] <SWPadnos> it was easy to find out, but not as easy as looking at a manual
[18:02:14] <SWPadnos> usrmot complains that it can't find an ini
[18:02:25] <SWPadnos> trying -i and -ini told me how to get it to load
[18:02:38] <SWPadnos> (looking at the source would have also, with some head scratching)
[18:02:44] <SWPadnos> then trying ? got me help :)
[18:02:49] <Lerman> I saw the message, but didn't conclude how to provide it.
[18:03:22] <SWPadnos> there should be (ought to be) some docs somewhere in the user interfaces section, I think
[18:03:33] <SWPadnos> BigJohnT, are you listening? :)
[18:03:36] <Lerman> show scales -> scales handled in HAL now!
[18:04:18] <SWPadnos> heh. my next question was what you were planning to do with usrmot
[18:04:36] <Lerman> From the integrator manual: The EMCMOT utility program USRMOT can be used to set these interactively and verify their results so that the proper values can be put in the INI file with a minimum of trouble.
[18:04:55] <SWPadnos> oh - thanks. that's old crufty user manual info
[18:04:58] <SWPadnos> ie, wrong
[18:05:22] <Lerman> There were apparently some (to me) arbitrary changes in the .ini file contents going from emc1 to emc2.
[18:05:41] <SWPadnos> I think there was a document on the wiki that told about the changes
[18:06:04] <Lerman> As in: my Y axis was backwards so that the feedback takes you in the wrong direction and causes an ferror.
[18:06:08] <SWPadnos> some are obvious (to me anyway), like removing the XXX_INDEX values - IO is in HAL now
[18:06:11] <Lerman> goody. let me go look.
[18:06:29] <SWPadnos> I don't know that it's there - I did a quick check earlier today
[18:07:15] <SWPadnos> ok - that's basic configuration knowledge (change the sign of the input or output scale, depending on which is wrong)
[18:07:23] <SWPadnos> that should be in the basic configuration section
[18:09:19] <SWPadnos> Lerman, please take some notes on what you don't find, and where you looked for it
[18:10:27] <Lerman> The key was knowing that was the problem. When I manually turned the Y axis, it jumped in the same direction that I turned it. That told me that feedback was in the wrong direction.
[18:11:49] <SWPadnos> ok
[18:12:18] <Lerman> It's tough to take notes when you are jumping back and forth from the office machine to the shop machine, but I'll try.
[18:12:31] <SWPadnos> I just noticed that the machine calibration program from the AXIS menu doesn't work (on sim/axis_mm anyway)
[18:12:34] <SWPadnos> heh
[18:12:50] <fenn> how are you going to calibrate a simulated machine
[18:13:17] <fenn> it's like a zen koan
[18:13:35] <SWPadnos> oh, true - I probably don't have PIDFF values in the ini - duh
[18:14:13] <fenn> if a cat steps on a keyboard and nobody is there to notice, does it make a sound?
[18:14:20] <SWPadnos> Lerman, if you can start EMC, then go to machine->Calibration in AXIS, that'll run a PID tuning dialog
[18:14:28] <SWPadnos> kind of like the old PIDTune, Ithink
[18:14:39] <SWPadnos> only if it's an IBM keyboard
[18:15:05] <SWPadnos> and not from a PCJr
[18:17:49] <SWPadnos> hey - I wonder where the program Fred mentions is?? http://www.linuxcnc.org/handbook/part2/pidtune.html
[18:24:04] <jepler> I dunno, no links to it in the thread and if there were they would probably be long dead. http://emc.sourceforge.net/NIST-archive/msg00489.html
[18:25:13] <SWPadnos> odd - I couldn't find any docs on how to "port" a config from EMC1 to EMC2
[18:25:54] <cradek> I don't recommend doing that
[18:26:19] <cradek> I recommend starting with a sample config and copying what you need. VERY little is useful to directly move over.
[18:26:33] <SWPadnos> even guidelines like "these are the names of the tools you use with EMC2 to do the things those other tools did with EMC1"
[18:27:17] <SWPadnos> and gotchas, like scaling for PID will probably be different due to the way HAL handles numbers (in engineering units)
[18:27:50] <anonimasu> SWPadnos: what units are that?
[18:27:53] <anonimasu> (just curious)
[18:31:54] <jepler> I don't remember emc1 well enough to recall, but it sounds like SWPadnos is implying that emc1 used something else (maybe feedback counts?) to find the error input to the PID algorithm.
[18:32:17] <Lerman> SWPadnos: Yup. As soon as calibration was mentioned, I went out to the shop to try running it and saw what it does.
[18:32:55] <SWPadnos> anonimasu, mm or inch - whatever you configure
[18:32:56] <Lerman> It might form a decent basis for an autotune program.
[18:33:10] <Lerman> I suspect that I have scaling problems.
[18:33:54] <cradek> do you have velocity mode amps?
[18:34:10] <Lerman> I have Jon Elson's stuff.
[18:34:30] <cradek> he has both - do you have tachometers?
[18:34:44] <Lerman> Which I would guess are torque mode. No tachs.
[18:36:01] <cradek> I'm not sure what 'correct scaling' would be then
[18:36:03] <Lerman> My input scaling seems right. I turn the handle one turn and the readout shows .200 inch.
[18:36:20] <cradek> I figured you were talking about pid input/output scaling
[18:36:46] <anonimasu> ah ok
[18:36:49] <Lerman> Yes. I am.
[18:38:16] <skunkworks> cradek: how did the resolver hookup work out?
[18:38:20] <cradek> what problem are you having?
[18:38:31] <cradek> skunkworks: I have working feedback on X so far
[18:38:37] <skunkworks> Nice :)
[18:38:50] <cradek> maybe tonight I'll move it - big step
[18:39:25] <skunkworks> your amps are +/- 10v?
[18:39:44] <cradek> yes (+- 8.5)
[18:39:48] <skunkworks> heh
[19:05:31] <Lerman> What is a reasonable value for MAX_ACCELERATION (small servos on a Bridgeport). My MAX_VELOCITY is 1.5 (inches per second).
[19:06:08] <SWPadnos> I've started out with some assumption about how fast they can get to full speed, like 0.2 seconds (giving you a max_accel 5x the max_vel)
[19:06:09] <cradek> 10?
[19:06:21] <Lerman> Is 6 inches/sec/sec/ too high?
[19:06:42] <Lerman> I guess the answer is no.
[19:06:43] <SWPadnos> that's a 0-full in 0.25, so I think it's fine (probably conservative)
[19:07:06] <cradek> does it follow well?
[19:07:09] <Lerman> I'm getting FERRORs.
[19:07:13] <SWPadnos> you can measure it with HAL, if you're so inclined
[19:07:20] <cradek> does the pid output saturate?
[19:07:23] <Lerman> I assume my tuning is way out of whack.
[19:07:29] <cradek> if so, you don't have the torque to accelerate that fast
[19:07:35] <cradek> if not, it's probably just mistuned
[19:07:40] <Lerman> I haven't looked yet.
[19:08:04] <SWPadnos> the old way of tuning the PPMC/USC was pretty weird - the output_scale numbers made no sense, so the PID numbers also made no sense
[19:08:26] <Lerman> So starting with the old parameters is a waste of time.
[19:08:36] <SWPadnos> that seems likely
[19:08:38] <cradek> my BOSS rapids at 4ips and I bet its acceleration is about 25ipss
[19:08:55] <SWPadnos> I don't know what torque mode parameters should look like though - it's not as straightforward as velocity mode
[19:08:56] <cradek> Lerman: proportionally the numbers are probably right
[19:09:09] <Lerman> The old value of P (3000) doesn't oscillate. When I push it resists, then let's go and ferrors.
[19:09:17] <SWPadnos> except for some square law stuff, due to the scale change
[19:09:21] <cradek> search for P (increase it until you barely get sustained oscillation) and then add all the rest in proportionally
[19:09:36] <Lerman> I'll go try that now. Thanks.
[19:09:44] <cradek> turn your ferror limits up
[19:09:50] <cradek> at least .5 inch
[19:09:57] <SWPadnos> yeah - you should tune with 6 inch FERROR ;)
[19:10:00] <cradek> otherwise you won't be able to see anything because it ferrors
[19:10:02] <SWPadnos> less on Z
[19:10:13] <cradek> SWPadnos: not 6" -- you want it to stop if it goes nuts
[19:10:17] <SWPadnos> heh
[19:10:22] <anonimasu> actually tach's and oscilloscope :p I'd advocate for tuning
[19:10:25] <SWPadnos> ok, 1" or so then
[19:10:42] <cradek> anonimasu: welcome to 1975
[19:10:47] <anonimasu> err setting acceleration :p
[19:10:56] <cradek> now we have halscope!
[19:11:13] <anonimasu> yep
[19:11:29] <anonimasu> that's what I meant :p
[19:11:53] <SWPadnos> halscope and ddt/vel encoder output ;)
[19:12:38] <SWPadnos> if the amps are torque mode, then theoretically, feeding a 1 into the DAC output should give 1 unit/sec^2 accel
[19:12:55] <SWPadnos> that might be the appropriate engineering unit to shoot for
[19:13:07] <SWPadnos> (not that it's easily measured)
[19:13:17] <anonimasu> :)
[19:13:20] <SWPadnos> or linear
[19:13:27] <SWPadnos> ot constant
[19:13:30] <SWPadnos> or
[19:13:58] <cradek> I don't think pid scaling matters for torque mode - it's just arbitrary
[19:14:22] <skunkworks> just tune it *tm
[19:15:09] <SWPadnos> yeah - I'm not sure how PID with position input can control torque (second derivative) well
[19:15:26] <SWPadnos> I bet FF1 or FF2 plays a big part
[19:15:34] <cradek> yes ff1 and ff2 are the important ones
[19:15:36] <skunkworks> that is how jone tuned it.
[19:16:11] <anonimasu> does anyone have a nice way of converting svf to dxf?
[19:16:12] <Lerman> Something is wonky. I just increased P to 200,000 and still no oscillation. I push it, it resists a little and then ferrs.
[19:16:26] <cradek> does it jump before it ferrors?
[19:16:40] <cradek> did you open up the ferror limits a bit?
[19:17:01] <cradek> it won't oscillate if the natural amplitude is higher than the ferror limit
[19:17:23] <SWPadnos> is this USC, UPC, or PPMC?
[19:17:27] <Lerman> I have FERROR = 0.001
[19:17:29] <Lerman> MIN_FERROR = .005
[19:17:42] <cradek> try MIN_FERROR = FERROR = .5
[19:17:42] <Lerman> The same as with EMC1.
[19:17:46] <SWPadnos> make that FERROR = 1, and comment out MIN_FERROR
[19:17:57] <SWPadnos> for tuning anyway
[19:18:03] <Lerman> OK. I'll go try that. Thanks.
[19:18:07] <cradek> you won't luck into following within .001!
[19:18:41] <cradek> * cradek sends Lerman some cat5
[19:18:51] <skunkworks> heh -
[19:18:57] <anonimasu> :)
[19:19:01] <SWPadnos> with P at 200000, PID is just about guaranteed to be saturating - I'm betting that the problem is downstream - like in the MIN/MAX_output or PID limits
[19:19:25] <skunkworks> or his following error is set to low. Oh wait..
[19:19:29] <cradek> I doubt there's a problem except it can't follow within .001 with a random tuning
[19:20:14] <anonimasu> yep
[19:20:14] <SWPadnos> sure, that's also got to be a problem
[19:20:42] <SWPadnos> it's also possible that the DAC scaling is off by either an order of magnitude (10V scale), or is inverted
[19:20:44] <cradek> if his bp is like mine, .001 is ten counts - that's a pretty tight tuning
[19:21:01] <cradek> he said it resists, so the feedback is the right way
[19:21:18] <SWPadnos> mathematical inverse, not negation
[19:21:28] <SWPadnos> 1/x
[19:21:40] <cradek> oh ok, we have changed that a few times haven't we
[19:21:48] <SWPadnos> I think so
[19:21:56] <alex_joni> 1/1/x ?
[19:22:05] <SWPadnos> x/1/1/x/x/1
[19:22:13] <SWPadnos> s/x/1/
[19:22:15] <SWPadnos> ;)
[19:22:21] <alex_joni> s/1/x/
[19:22:31] <SWPadnos> s/1/x/g
[19:22:32] <cradek> wonder what accel is on the BOSS. it sure gets up to rapid speed fast.
[19:22:46] <SWPadnos> servos or steppers?
[19:22:50] <alex_joni> s/accel/BOSS/
[19:22:51] <cradek> servos
[19:22:57] <alex_joni> s/BOSS/foo/
[19:23:02] <SWPadnos> probably gets to speed in ~0.1-0.2 seconds
[19:23:23] <cradek> that would be 20-40 ipss
[19:23:24] <SWPadnos> (seat of the pants estimate - I have no experience with those machines)
[19:23:34] <cradek> yeah it's in there somewhere I bet.
[19:23:44] <SWPadnos> a lot of servos will accel to full speed in much less than a single turn
[19:23:45] <cradek> maybe even 50.
[19:24:32] <cradek> before I pull out that control I could scope the tach and (maybe) see what the design spec was
[19:26:35] <SWPadnos> indeed
[19:27:09] <sculptor> hi!
[19:27:14] <cradek> hello
[19:27:30] <sculptor> does this channel cover CNC machine maintenance too
[19:27:53] <anonimasu> yes
[19:28:22] <sculptor> brilliant
[19:28:47] <SWPadnos> are you vrsculptor from the Yahoo CAD CAM EDM DRO group?
[19:29:05] <sculptor> no. i found this channel by using google
[19:29:10] <SWPadnos> ah, ok
[19:29:13] <sculptor> i am not vrsculptor
[19:30:02] <sculptor> this is my modest contribution to the CNC community : http://www.machinetoolhelp.com/forum/viewtopic.php?id=200
[19:30:16] <sculptor> a program that can crack fanuc ladder logic passwords
[19:30:23] <sculptor> anyway
[19:30:42] <SWPadnos> heh
[19:30:46] <sculptor> i have a backslah in one ball screw, on a EMAG VL5 macine
[19:31:20] <sculptor> x axis. it seems that the the ball screw is multithreaded - but only one track is loaded with balls
[19:31:27] <SWPadnos> hmmm. are you a member of that Yahoo group? you may be able to get specific information there
[19:31:28] <sculptor> i plan to use the other track!
[19:31:37] <sculptor> but
[19:31:57] <sculptor> a nut is probably damaged too. i'll have it measured on a 3d machine, tomorrow
[19:31:57] <SWPadnos> I'm not sure the folks here have as much specific machine experience
[19:32:01] <Lerman> OK. I set P to 100, hit the edge of the wheel with the heel of my hand and it starts oscillating at (I guess) a few cycles per second.
[19:32:22] <sculptor> i need larger balls, no pun intended!
[19:32:27] <jepler> move
[19:32:29] <Lerman> Set it at 95 and it oscillates briefly and damps out.
[19:32:30] <jepler> argh
[19:32:38] <sculptor> how to calculate diameter for new balls
[19:32:38] <jepler> * jepler is using eagle today, but keeps focusing the wrong window
[19:32:42] <SWPadnos> Lerman, you may want to turn DEADBAND down to 0 also, it may prevent oscillation
[19:32:47] <cradek> ok, leave it at 100, add a little D and it will stop
[19:33:14] <Lerman> You think 100. Or should I turn it down to 95?
[19:33:28] <cradek> then (leaving I at 0) work on ff1,ff2 until a jog has good following, then turn up I
[19:33:35] <Lerman> Deadband is 0.00005
[19:33:35] <cradek> leave it at 100 is fine
[19:33:39] <anonimasu> sculptor: I havent done anything like that, but it seems that most people that reload ballscrews at home do it by trial and error
[19:33:53] <cradek> that sounds right if your encoders give you .0001 like mine do
[19:34:21] <Lerman> I'll head back to the shop. Input scale is 20,000.
[19:34:50] <cradek> try D of 1-5 or so, FF1,FF2 will be even lower
[19:34:50] <Lerman> So the encoders give .00005 == the deadband.
[19:35:00] <Lerman> BBL.
[19:36:01] <sculptor> i'm not doing that at home, though :) i'm doing it at work, and i have loads of resources - i can egalize nut's track by grinding it, for example
[19:36:17] <sculptor> i just need instructions :)
[19:36:23] <sculptor> knowledge, experience
[19:36:43] <anonimasu> sorry then I have no idea :)
[19:37:10] <cradek> I bet most send them to rebuilders, it will be hard to find someone who has done a lot of them
[19:37:18] <anonimasu> yep
[19:38:07] <sculptor> i have very limited choice of rebuilders in this area, and i don't trust easily to anyone :) i want to learn something, too
[19:38:18] <sculptor> btw, i already ordered new ballscrew
[19:38:26] <cradek> :-)
[19:38:32] <cradek> that makes it a lot more fun I bet
[19:38:32] <sculptor> if it's broken already, you cannot broke it!
[19:38:41] <sculptor> break*
[19:39:12] <sculptor> anyway, SWPadnos mentiones a jahoo group: CAD CAM EDM DRO
[19:39:15] <sculptor> mentioned*
[19:39:25] <sculptor> do they cover CNC machine maintenance
[19:39:35] <sculptor> i need everything : both mechanics and electronics
[19:39:50] <sculptor> hydraulics, pneumatics, etc
[19:39:56] <SWPadnos> well, yes and no - they're about building CNC machines, or retrofitting manual machines
[19:40:01] <sculptor> i have 100+ CNC machines at work
[19:40:14] <SWPadnos> but if you ask about regrindding a screw or figuring out what size to re-ball it, they'll probably help out
[19:40:15] <Lerman> Ahha. I added D of 1. I was then able to turn P up to 250 which makes it 2-1/2 times as stiff. At 300, I get oscillation.
[19:40:16] <sculptor> and i could really use someone's experience
[19:40:45] <Lerman> Does it make sense to keep iterating. Increase D, again and see how far I can push P?
[19:40:51] <SWPadnos> I think the CCED group is probably your best bet. there are other groups as well, but I don't know them offhand
[19:40:57] <cradek> Lerman: you are on the right track. D is touchy though, eventually it won't be stable
[19:41:10] <sculptor> loads of EMAGs, Dawoos, Mazaks and MoriSeikia
[19:41:11] <sculptor> s
[19:42:19] <sculptor> CNCs rock!
[19:42:21] <anonimasu> :)
[19:42:35] <SWPadnos> Stuart Stevenson is probably your best bet among the people on this channel (though he's not on right now)
[19:42:37] <cradek> I like to make a loop that does alternating short rapids (1in) with a 1sec pause in between, then carefully set scope to trigger, and you can tweak while watching it
[19:42:45] <skunkworks> sculptor: have you used emc?
[19:42:49] <sculptor> no
[19:42:55] <sculptor> is that a come sort of controller
[19:42:58] <sculptor> some*
[19:43:04] <SWPadnos> yes
[19:43:05] <sculptor> NC controler?
[19:43:06] <BigJohnT> SWPadnos: I'm back
[19:43:07] <cradek> free/open source pc-based
[19:43:11] <sculptor> ah, software
[19:43:15] <SWPadnos> enhanced machine controller, started at NIST
[19:43:21] <SWPadnos> yay!
[19:43:21] <jepler> move
[19:43:25] <jepler> argh I did it again
[19:43:28] <SWPadnos> please write the missing manual sections ;)
[19:43:28] <cradek> rats!
[19:43:30] <skunkworks> jepler: stop it ;)
[19:43:36] <sculptor> i know about Mach2
[19:43:40] <jepler> "think to focus" is really buggy today
[19:43:42] <BigJohnT> ok, which ones?
[19:43:42] <SWPadnos> ewwww
[19:43:43] <sculptor> now i know about emc too
[19:44:15] <sculptor> if i am about to build a cnc machine
[19:44:30] <SWPadnos> oh right - lerman was looking for docs on usrmot, and there aren't any
[19:44:30] <sculptor> i would build controller from scratch!
[19:44:35] <Lerman> So, I'll push. Is there a textual way (I'd think so) to set up halscope? Anyone have a file I can use as a start.
[19:44:41] <cradek> what is usrmot?
[19:44:53] <skunkworks> sounds like something from emc1
[19:44:54] <sculptor> i'm familiar with bresenham's algorithms, etc
[19:45:03] <anonimasu> sculptor: it gets a little hard when you need to deal with the trajectory planning and stuff :)
[19:45:04] <SWPadnos> like halcmd for EMC-type commands (move in a circle, turn on coolant ...)
[19:45:07] <Lerman> There is a reference to it in the Integrator Manual.
[19:45:13] <cradek> eek
[19:45:20] <cradek> BigJohnT to the rescue?
[19:45:20] <sculptor> heh, i bet
[19:45:38] <BigJohnT> * BigJohnT swoops down and crashes into a tree
[19:45:44] <cradek> ha
[19:45:46] <SWPadnos> Lerman, there's no need to "start from a config" with halscope - you'll want to connect the channels as you wish, and a .scope.cfg file will be saved in the current dir when you exit
[19:45:53] <sculptor> so, how do you connect a machine to a servo driver? parallel port?
[19:45:54] <Lerman> The suggestion in the manual is that it could help with setting tuning parameters.
[19:45:58] <sculptor> s/machine/PC/
[19:46:09] <SWPadnos> maybe remove the reference in the manual ;)
[19:46:14] <cradek> sculptor: lots of options from parallel port up to analog servo controls
[19:46:20] <SWPadnos> since it isn't useful for that any more
[19:46:21] <Lerman> OK. I didn't know it would save things. I started to use it, exited and then didn't feel like restarting it.
[19:46:24] <anonimasu> sculptor: lots of options.. servo driver cards/step generators or paralell port
[19:46:39] <SWPadnos> gotta be in the same dir though, or copy .scope.cfg
[19:46:40] <sculptor> brilliant
[19:46:44] <Lerman> Change the reference in the manual to be a reference to Calibration in Axis!
[19:46:49] <SWPadnos> indeed
[19:46:55] <SWPadnos> gotta run -see you later
[19:46:58] <Lerman> I'm back to the shop. Thanks for all your help.
[19:47:02] <anonimasu> sculptor: if you are interested you should look at youtube for some videos of emc in action :)
[19:47:13] <sculptor> way ahead of you, anonimasu !
[19:47:14] <BigJohnT> was USRMOT replaced with the servo configure in axis?
[19:47:27] <anonimasu> and http://www.anderswallin.net
[19:47:29] <BigJohnT> yep calibration
[19:47:36] <anonimasu> that one's got some 3d contouring
[19:47:38] <anonimasu> :
[19:47:39] <anonimasu> :)
[19:48:17] <BigJohnT> Lerman: I actually have some screen shots of servo configure and was working it into the manual :)
[19:49:03] <BigJohnT> servo calibration I meant
[19:50:01] <sculptor> does emc support backslash cancelation
[19:50:09] <sculptor> can you implement m codes tooo
[19:50:29] <sculptor> s/backslash/backlash/
[19:50:43] <sculptor> s/cancelation/cancellation/
[19:50:48] <sculptor> damn, i'm a spell nazi
[19:53:04] <skunkworks> yes, yes
[19:53:14] <sculptor> gut, gut
[19:53:36] <skunkworks> it can also do leadscrew mapping
[19:54:02] <skunkworks> or - I should say - you can have emc adjust for leadscrew non-linearity
[19:55:48] <sculptor> hm
[19:56:02] <sculptor> by mapping - you mean positional error correction
[19:56:47] <sculptor> 90% of my machines have double feedback devices - a rotary incremental encoder on axis' motor and an absolute linear encoder
[19:58:07] <sculptor> sipindles have contactless gear incremental encoder
[20:03:03] <anonimasu> sculptor: that's probably not something anyone have done yet :)
[20:03:31] <anonimasu> sculptor: that kind of machines arent showing up with broken controls just yet :)
[20:05:32] <cradek> yes we've done a machine with double feedback - it worked great
[20:05:43] <anonimasu> really?
[20:05:45] <anonimasu> neat!
[20:05:47] <cradek> yes
[20:07:08] <anonimasu> im still waiting for encoders for my machine -_-
[20:07:12] <cradek> well there might be others out there, but that was the first one I worked on personally
[20:07:12] <anonimasu> I cant afford them this month
[20:09:13] <cradek> http://article.gmane.org/gmane.linux.distributions.emc.user/8424
[20:11:32] <anonimasu> nice
[20:11:55] <cradek> that green 5-axis cincinatti is also running EMC2 now
[20:14:15] <alex_joni> good night all
[20:14:27] <cradek> g'night
[20:14:44] <anonimasu> night alex
[20:19:28] <sculptor> heh
[20:20:04] <sculptor> s/axis/axle/
[20:20:56] <sculptor> older machines have linear incremental encoder + motor based tachogenerator
[20:21:27] <cradek> yes, velocity mode servo amps
[20:21:51] <sculptor> encoder is connected to NC. NC controlls a drive via analog velocity signal
[20:21:59] <cradek> even older, tachometer + resolver, like the machine I'm working on retrofitting with EMC2 control
[20:23:30] <sculptor> such machines also have some sort of 3 output encoder
[20:23:41] <anonimasu> resovers..
[20:23:42] <sculptor> used mostly for direction detection, iirc
[20:23:43] <sculptor> ah, ok
[20:23:52] <sculptor> i only have 2 of those
[20:23:58] <cradek> tachometer is velocity sensor, resolver is position sensor
[20:24:02] <sculptor> aye
[20:24:15] <cradek> no encoders (unfortunately)
[20:25:13] <sculptor> those two machines have indramat controller and drivers
[20:25:28] <sculptor> indramat does not exist anymore (bought by bosch rexroth)
[20:25:51] <sculptor> bosch rexroth is a great company! hydraulics, pneumatics, electronics, everything
[20:26:56] <anonimasu> yep, though they also make you pay for it
[20:26:58] <anonimasu> :)
[20:28:07] <sculptor> well, i fixed several of those modules so far
[20:28:12] <sculptor> haven't bought a single one
[20:28:25] <sculptor> last one had a problem with excessive courrent
[20:28:36] <sculptor> i opened it up, and found that a hall sensor for current detection is faulty
[20:28:41] <sculptor> so, i ordered new one
[20:28:46] <sculptor> that did the trick
[20:28:50] <sculptor> $20, baby
[20:28:52] <anonimasu> I've dealt alot with their huydralic stuff
[20:29:02] <sculptor> me too
[20:29:12] <sculptor> we use PV7 vane pumps a lot
[20:29:30] <sculptor> btw, i'm a chief of 15 men strong maintenance department
[20:29:51] <anonimasu> ah I see
[20:29:53] <anonimasu> nice
[20:30:23] <sculptor> coming from EE, i like to teach my mechanical engineers and technicians about mechanics!
[20:30:54] <sculptor> i have only 11 months of experience in CNCs, though
[20:31:20] <BigJohnT> hi CanSir
[20:32:13] <anonimasu> this is disturbing.. i
[20:32:22] <sculptor> hydraulics is my favorite area
[20:32:29] <anonimasu> it seems like inkscape wont export dxf files...
[20:32:37] <CanSir> Hi BigJohn !
[20:32:51] <BigJohnT> you get it working?
[20:32:54] <sculptor> everyone in my department sucks in hydraulics diagnostics, but me!
[20:33:14] <CanSir> Yes thanks. I've even made some progress on my auto tool length probe.
[20:33:22] <BigJohnT> cool
[20:33:22] <sculptor> hm
[20:33:42] <anonimasu> sculptor: and that's probably the easiest stuff to find the problems with too :)
[20:33:48] <sculptor> we call those 'tool break detection"
[20:33:54] <sculptor> aye, anonimasu
[20:34:04] <sculptor> everything's logical, if you can read schematics
[20:34:13] <sculptor> well, you cannot measure flows, only pressures
[20:34:18] <sculptor> that makes it more interesting
[20:34:19] <CanSir> My machine is too wimpy to break the tool. :)
[20:34:20] <anonimasu> you can with a flow meter
[20:34:26] <BigJohnT> * BigJohnT is finished rotating the inserts in this 3" face cutter and goes back to work
[20:34:40] <sculptor> yes, but in order to use one, you have to break a circuit
[20:34:49] <sculptor> and expose it to dirt containtment
[20:34:55] <sculptor> contamination, even
[20:35:16] <sculptor> CanSir, heh
[20:35:42] <sculptor> machines at work are strong enough to move 100kg+ clamping tool
[20:35:57] <sculptor> or to break 2cm diameter tool
[20:36:25] <CanSir> I'm more in with the Dremel crowd. :o
[20:36:33] <CanSir> :D
[20:38:12] <anonimasu> sculptor: you have the same issues when you assemble stuff too.. with contamination..
[20:38:25] <sculptor> how do you call a part that encloses machining area, and protect the machine from chip containment - a curtain?
[20:38:32] <sculptor> (in english)
[20:38:35] <anonimasu> encolsure
[20:38:48] <anonimasu> err enclosure :p
[20:38:56] <sculptor> hm, no
[20:39:05] <sculptor> i'm not talking about a door, or a box
[20:39:37] <anonimasu> well something that encloses the machine is a enclosure..
[20:39:57] <CanSir> Are you talking about a shoe? It covers the bit and usually has a vacuum port on it?
[20:40:08] <sculptor> i'm talking about a set of metal bands, connected together
[20:40:22] <anonimasu> way covers or chip guards
[20:40:32] <sculptor> there you go
[20:42:49] <sculptor> very often, workers make mistake when entering tool offset, they run a program without limiting traverse, a tool hits a clamping tool, clamping tool jumps out of its nests, and falls on a chip guard, and bends it totally
[20:43:05] <sculptor> nest
[20:43:51] <sculptor> then, my technicians have to return the clamping tool to its position, to take apart chip guard, and try to flatten it
[20:44:03] <sculptor> they are rather expensive - cheapest ones are 3k eur
[20:44:31] <anonimasu> big/long ones right?
[20:44:38] <sculptor> aye
[20:45:14] <sculptor> with wipers, and small rollers (608 bearings actually)
[20:50:28] <sculptor> unless sorrounded by a "
[20:50:32] <sculptor> su
[20:50:50] <sculptor> lol, wrong chan
[21:13:55] <CanSir> On some of my test configurations of EMC2 I see a "Home Axis" button in AXIS, but on my real configuration I see a "Home All" button. Can anybody explain how I managed that?
[21:14:27] <BigJohnT> home all is when you have the homing sequence defined for each axis in your ini file
[21:15:00] <CanSir> Ah. I must have set that up by mistake. I don't have any home/limit switches installed yet.
[21:15:53] <BigJohnT> look for HOME_SEQUENCE in your ini file
[21:16:19] <CanSir> Will do - thanks yet again BigJohnT !
[21:16:40] <BigJohnT> and in Configuration/INI configuration/homing in the Integrators manual
[21:16:43] <BigJohnT> np
[21:18:41] <BigJohnT> CanSir: did you use stepconf on your "real" config?
[21:22:47] <CanSir> Initially I did - yes.
[21:25:04] <BigJohnT> if you didn't modify it after you can just run stepconf again and put not used for the switches
[22:39:28] <MASEngr> Hi everyone. I've had a very interesting day, and I require help getting the joypad to work properly. There were some hardware issues, and long story short, the newly rebuilt machine is possibly working. When I try to start up emc2, it crashes with a "pin does not exist" error.
[22:39:29] <MASEngr> http://pastebin.com/m77608eb7
[22:40:23] <MASEngr> I've had that before once, amusingly due to a "not plugged in" error. That's not the case this time, as I can see the joypad as /dev/input/js0 and calibrate it.
[22:41:38] <MASEngr> Is there anything I can do to get this mill up and running? I've had a rough day, so excuse any typos.
[22:43:44] <MASEngr> Can someone reply with a "hello MASEngr" so I know that at least XChat is working? It's been that kind of day. Thank you.
[22:44:53] <skunkworks> hello MASEngr :)
[22:45:20] <MASEngr> Oh good, thanks, skunkworks. ;)
[22:45:30] <skunkworks> can you pastebin your hal files?
[22:45:41] <MASEngr> But of course!
[22:45:54] <skunkworks> * skunkworks might not be able to help... but someone should be able to
[22:46:38] <MASEngr> http://pastebin.com/d6f89fbab
[22:47:09] <MASEngr> http://pastebin.com/d44db8a0b
[22:47:36] <MASEngr> http://pastebin.com/d155f649c
[22:47:54] <MASEngr> http://pastebin.com/d487fab92
[22:48:14] <MASEngr> http://pastebin.com/d4cacb673
[22:49:23] <MASEngr> There we go. This was working fine until the incident with the hard drive and the upgrading and the Very Long Days. I got the hal files off the old machine's dead drive and got them back onto the new machine.
[22:50:14] <MASEngr> Which means I'm using the Hardy Heron CD - The Gutsy Gibbon version will not boot.
[22:52:58] <skunkworks> that is good - as emc is not officailly supported on gutsy gibbon
[22:53:48] <skunkworks> according to this http://www.linuxcnc.org/docs/2.2/html/man/man1/hal_joystick.1.html
[22:54:25] <MASEngr> Er, sorry, Dapper Drake. I mis-typed.
[22:54:35] <skunkworks> oops - never mind. (thought I had it)
[22:54:56] <skunkworks> sorry
[22:57:32] <MASEngr> That's all right. I can see that the syntax is correct, and the parameters are what the rest of the file expects.
[22:57:54] <skunkworks> I think your going to have to do a loadusr hal_joystick -d /dev/input/js0 -p joypad
[22:58:06] <MASEngr> In joypad.hal, I have loadusr hal_joystick -d /dev/input/js0 -p joypad
[22:58:14] <skunkworks> manually and see what pins show up. It looks correct as far as I can tell
[22:59:41] <MASEngr> RTAPI: ERROR: could not open shared memory (errno=2)
[22:59:41] <MASEngr> HAL: ERROR: rtapi init failed
[22:59:41] <MASEngr> halcmd: hal_init() failed: -9
[22:59:41] <MASEngr> NOTE: 'rtapi' kernel module must be loaded
[23:00:23] <skunkworks> oh - your at your computer?
[23:00:25] <skunkworks> heh =
[23:01:03] <MASEngr> Yep, and having it up and running is a Big Step from yesterday.
[23:01:22] <skunkworks> can someone explain how to run halcmd with installed emc?
[23:01:22] <MASEngr> I'm not sure what that error message is supposed to tell me.
[23:03:03] <skunkworks> let me get a linux machine booted
[23:03:43] <skunkworks> this is a nice hal tutorial http://www.linuxcnc.org/docs/2.2/html/hal_tutorial.html
[23:04:00] <skunkworks> but it assumes a RIP install - I am assuming you installed the emc live cd?
[23:05:03] <CanSir> How can I see what is sent to STDOUT within EMC? Example g-code: (print, This is a test)
[23:06:45] <skunkworks> MASEngr: do you have a terminal open?
[23:06:52] <MASEngr> Hmm, odd, when I run the commands from the halcmd line, it doesn't return errors.
[23:07:00] <MASEngr> I've always got a terminal window open.
[23:07:03] <MASEngr> ;)
[23:07:03] <skunkworks> heh
[23:07:44] <MASEngr> Yes, I installed with the Live CD for Ubuntu Hardy Heron, or whatever the "most recent" version is, burned last night.
[23:07:51] <skunkworks> hehhal_joystick -d /dev/input/js0 -p joypad
[23:08:15] <skunkworks> oops - did you run halcmd - then entered hal_joystick -d /dev/input/js0 -p joypad
[23:08:42] <MASEngr> Yes, I ran that with the "loadusr" before it.
[23:08:47] <skunkworks> right
[23:09:10] <skunkworks> ok - can you pastbin what gets output with 'show'?
[23:09:23] <MASEngr> From the line, I didn't get an error, and the second time (FSAG) I got:
[23:09:23] <MASEngr> halcmd: loadusr hal_joystick -d /dev/input/js0 -p joypad
[23:09:24] <MASEngr> halcmd: HAL: ERROR: duplicate variable 'joypad.button.0'
[23:09:24] <MASEngr> ERROR: button 0 export failed with err=-3
[23:09:48] <skunkworks> you can'
[23:09:55] <skunkworks> you can't load it twice
[23:10:19] <skunkworks> type show and pastbin the results
[23:10:25] <MASEngr> I know, I was hoping for a helpful error message. The fact that it loaded the first time was pretty good.
[23:10:38] <MASEngr> Hang on, I'm apt-getting "show".
[23:11:04] <skunkworks> no - in halcmd - type the command show..
[23:11:40] <skunkworks> that show all the pins and stuff that are available at this time.
[23:11:55] <skunkworks> stuff is a technical term
[23:12:16] <MASEngr> Sure, wait a moment. Recovering from PEBKAS error.
[23:13:38] <MASEngr> http://pastebin.com/d66ffae3b
[23:15:33] <MASEngr> And if I hold axes / press buttons, that's refected in "show". Very weird.
[23:15:48] <skunkworks> well - I don't know - the pins that are available are what the hal file is using. I think you may have to wait until someone with more knowledge is here. (every one seems to be hiding) ;)
[23:16:01] <skunkworks> cool huh?
[23:16:55] <skunkworks> joypad.axis.0 looks correct and that is how it is in your hal file that is causing the error net velX joypad.axis.0 => sim-encoder.0.speed
[23:17:08] <MASEngr> Yeah, that's the confounding part. I can see the device, I know that it's there, and the system is responding correctly.
[23:17:21] <MASEngr> The script was working, and now it just crashes when it boots emc2.
[23:18:10] <skunkworks> joypad.hal:17: pin 'joypad.axis.0' does not exist
[23:18:18] <skunkworks> weird
[23:18:31] <MASEngr> I appreciate the help nevertheless, and hey, at least I got Xchat working. ;)
[23:18:49] <skunkworks> heh
[23:19:05] <MASEngr> I spent the better part of 4 hours getting the joystick working, period.
[23:19:31] <MASEngr> It just wasn't showing up in the lsusb at all. It turned out to be a BIOS error.
[23:19:37] <skunkworks> yeck
[23:19:56] <MASEngr> Well, in all fairness, it wasn't the only problem, but it was the biggest one.
[23:20:20] <MASEngr> I've been struggling with this since early this morning, when I finally got the old config files from the older machine.
[23:20:42] <MASEngr> Would the scripts change between kernel versions?
[23:21:03] <MASEngr> I can't see how, given that the command works from the command line but not the script.
[23:21:49] <skunkworks> for shits and grins - do a loadrt sim_encoder num_chan=3
[23:21:59] <MASEngr> Yeah, that worked too.
[23:22:00] <skunkworks> and then a net velX joypad.axis.0 => sim-encoder.0.speed
[23:22:41] <MASEngr> I got that far without errors. Maddening, I tell you, maddening.
[23:23:02] <skunkworks> the net velX joypad.axis.0 => sim-encoder.0.speed worked?
[23:24:05] <skunkworks> MASEngr: net velX joypad.axis.0 => sim-encoder.0.speed worked?
[23:24:48] <MASEngr> halcmd: loadusr hal_joystick -d /dev/input/js0 -p joypad
[23:24:48] <MASEngr> halcmd: loadrt encoder num_chan=3
[23:24:48] <MASEngr> halcmd: loadrt sim_encoder numchan=3
[23:24:48] <MASEngr> insmod: error inserting '/usr/realtime-2.6.24-16-rtai/modules/emc2/sim_encoder.ko': -1 Unknown symbol in module
[23:24:48] <MASEngr> <stdin>:3: exit value: 1
[23:24:50] <MASEngr> <stdin>:3: insmod failed, returned -1
[23:24:52] <MASEngr> halcmd: loadrt sim_encoder num_chan=3
[23:24:54] <MASEngr> halcmd: net velX joypad.axis.0 =>sim-encoder.0.speed
[23:24:56] <MASEngr> halcmd: net velY joypad.axis.1 =>sim-encoder.1.speed
[23:24:58] <MASEngr> halcmd: net velZ joypad.axis.2 =>sim-encoder.2.speed
[23:25:08] <MASEngr> Pardon my butterfingers on the first sim_encoder load, but yep, it works.
[23:26:13] <MASEngr> (I typed numchan instead of num_chan)
[23:26:54] <MASEngr> http://pastebin.com/db5e0cfc
[23:28:10] <MASEngr> The best part of that is how it's clearly mapping correctly.
[23:28:25] <MASEngr> #
[23:28:25] <MASEngr> Signals:
[23:28:25] <MASEngr> #
[23:28:25] <MASEngr> Type Value Name (linked to)
[23:28:25] <MASEngr> #
[23:28:26] <MASEngr> float 0 velX
[23:28:28] <MASEngr> #
[23:28:30] <MASEngr> <== joypad.axis.0
[23:28:32] <MASEngr> #
[23:28:34] <MASEngr> float 0 velY
[23:28:36] <MASEngr> #
[23:28:40] <MASEngr> <== joypad.axis.1
[23:28:42] <MASEngr> #
[23:28:44] <MASEngr> float 0 velZ
[23:28:46] <MASEngr> #
[23:28:48] <MASEngr> <== joypad.axis.2
[23:30:05] <skunkworks> wierd
[23:30:05] <MASEngr> I mean no offence, but do you know what time zone Alex is in?
[23:30:19] <skunkworks> he is in ro
[23:30:25] <skunkworks> I would guess sleeping.
[23:30:26] <skunkworks> :)
[23:30:42] <MASEngr> Oh well.
[23:31:06] <MASEngr> So you agree that this should be working, right? I'm not crazy?
[23:31:15] <skunkworks> sorry - I really don't know why it wouldn't be working. You just did it manually and it worked.
[23:31:38] <skunkworks> it is the joypad.hal that is erroring.. and that is what is in it. odd
[23:32:54] <skunkworks> * skunkworks looks around for some help ;)
[23:33:59] <skunkworks> you could # out all the lines in the joypad hal file except for the loadusr hal_joystick -d /dev/input/js0 -p joypad line so you could see within emc what is happening..
[23:36:12] <MASEngr> Hmm, interesting:
[23:36:13] <MASEngr> Debug file information:
[23:36:13] <MASEngr> insmod: error inserting '/usr/realtime-2.6.24-16-rtai/modules/rtai_hal.ko': -1 File exists
[23:36:13] <MASEngr> 8368
[23:37:25] <skunkworks> reltime is probaly already running.. (from halcmd) - did you exit out of it?
[23:38:03] <skunkworks> realtime
[23:40:12] <MASEngr> Ignore the last thing I said - I forgot to comment out one of the lines. Now I've got emc2 up and running, albeit without some of the parameters.
[23:40:38] <MASEngr> Whatever was in joypad.hal, of course.
[23:42:26] <MASEngr> Well, let's power up the mill and see if I can get it to start up. That would be super.
[23:43:31] <MASEngr> Well, I have spindle control, so that's something.
[23:43:39] <skunkworks> heh
[23:43:48] <MASEngr> This whole problem is weirder than a plaid suit and a banana hat.
[23:44:08] <skunkworks> is this the mill that has scales and used to have an anilam control on it?
[23:44:22] <jmkasunich_> MASEngr: regarding the joypad thing
[23:44:25] <MASEngr> If I say yes, are you going to stop helping?
[23:44:37] <skunkworks> Heh - no
[23:44:43] <jmkasunich_> I haven't read your hal files yet - just skimmed thru all of today's IRC log
[23:44:50] <jmkasunich_> but I have a theory
[23:45:04] <MASEngr> Okay, jmk, I'll try it.
[23:45:23] <jmkasunich_> when you loadusr joypad manually, there is plenty of time between that and the next command, whether it is a "show" or a link
[23:45:28] <MASEngr> But yes, skunkworks, this is that mill.
[23:45:47] <MASEngr> Oh FFS, of course. THe new machine is 100x faster.
[23:46:05] <jmkasunich_> but in your HAL script, the next command happens very quickly - joypad might not have had time to create the HAL pins yet, so when you try to link to them they are missing
[23:46:12] <MASEngr> Literally - max jitter has gone from 800k to 8k.
[23:46:28] <MASEngr> Okay, so can I pause in the script?
[23:46:35] <jmkasunich_> loadusr has options that tell it to wait until the component you loaded is ready
[23:46:51] <jmkasunich_> try the halcmd man page for details, I don't have it at the top of my head
[23:47:06] <skunkworks> oh - very interesting.. I would have not known.
[23:48:02] <skunkworks> wow - max jitter of 800k is not good for really - anything.
[23:49:18] <MASEngr> Yeah, which was sort of the point of the upgrade.
[23:49:41] <MASEngr> Does -W flag ring a bell, so the command should be loadusr -W hal_joystick -d /dev/input/js0 -p joypad
[23:50:53] <skunkworks> looks like -W
[23:51:34] <MASEngr> Weird, now it's taking forever to load, and a restart gives an error:
[23:52:05] <MASEngr> Debug file information:
[23:52:05] <MASEngr> 9401
[23:52:05] <MASEngr> Stopping realtime threads
[23:52:05] <MASEngr> Unloading hal components
[23:52:06] <MASEngr> /usr/bin/emc: line 356: emc.nml: No such file or directory
[23:52:08] <MASEngr> Waiting for component 'hal_joystick' to become ready......................................................................................................................................................................................................................................................................................................................................................................................................
[23:52:15] <MASEngr> ......................................
[23:52:17] <MASEngr> joypad.hal:10: hal_joystick exited without becoming ready
[23:52:19] <MASEngr> Stopping realtime threads
[23:52:21] <MASEngr> Unloading hal components
[23:52:23] <MASEngr> ERROR: Module hal_lib does not exist in /proc/modules
[23:52:25] <MASEngr> ERROR: Module rtapi does not exist in /proc/modules
[23:52:27] <MASEngr> ERROR: Module rtai_math does not exist in /proc/modules
[23:52:29] <MASEngr> ERROR: Module rtai_sem does not exist in /proc/modules
[23:52:31] <MASEngr> ERROR: Module rtai_fifos does not exist in /proc/modules
[23:52:33] <MASEngr> ERROR: Removing 'rtai_sched': Device or resource busy
[23:52:35] <MASEngr> ERROR: Module rtai_hal is in use by rtai_sched
[23:52:37] <MASEngr> Wait a moment...
[23:52:48] <skunkworks> anything more than a few lines - pastbin
[23:52:54] <skunkworks> pastebin
[23:53:19] <MASEngr> Yeah, that was more info than I thought, sorry about that. At least I didn't paste the whole error log. :{
[23:53:30] <skunkworks> heh
[23:53:45] <skunkworks> we would have reached thru the computer and strangled you.
[23:53:49] <MASEngr> core_servo.hal shows the thread period for the joypad as:
[23:53:56] <MASEngr> loadrt threads name1=joy-thread fp1=0 period1=10000000
[23:54:26] <MASEngr> Isn't that on best of bash? "Someday I'll become rich and famous after I invent a way to stab people over the internet"
[23:59:52] <MASEngr> I tried cutting the period to 100000, failed. Trying 1000000.