#emc | Logs for 2009-08-01

[02:18:12] <Jymmm> If you have an iPhone update it now!
[10:40:31] <El_Matarife> Anyone going to the North Texas CNC user group meeting today?
[10:43:21] <GammaX> would anyone like to make some extra money making me a fitting?
[10:45:26] <kanzure> El_Matarife: No, but I would have liked to have known about it sooner. :-)
[10:45:59] <El_Matarife> Well I didn't find out about you guys till today when I was googling "CNC IRC"
[10:46:09] <El_Matarife> It's been in the CNC zone thread for Dallas for a while
[10:46:26] <El_Matarife> http://www.ntcncug.com/index.htm
[10:46:44] <El_Matarife> Joe from JoesCNC is showing up today apparently
[10:47:08] <GammaX> or I should ask if anyone kows where to buy an encoder housing for sem servos?
[10:47:16] <El_Matarife> I'm currently 75% done with Joe's 4x4 hybrid and I'm considering this instead of Mach
[11:18:33] <SWPadnos> GammaX, don't buy encoders yet, bonehead! :)
[11:18:53] <GammaX> didnt u tell me to buy encoders lol
[11:19:24] <GammaX> I have old ass glass scales from the 80's haha
[11:20:08] <SWPadnos> encoders might help, since EMC would then have feedback from both the motor and the table
[11:20:34] <SWPadnos> but then cradek pointed out that replacing things willy-nilly isn't a good way to debug, and I agree with that
[11:21:03] <GammaX> I know but either way I will need encoders.
[11:21:20] <SWPadnos> if one or more scales are broken, then it is probably easier to fit encoders to the motors instead of fixing or replacing the scales
[11:21:45] <GammaX> SWPadnos, the scales work properly and are giving correct feedback to emc.
[11:21:52] <SWPadnos> that may not be true. I think if backlash is low enough you can get away without them
[11:22:41] <GammaX> its an old machine and I dont know how to properly check backlash
[11:23:28] <GammaX> SWPadnos, Id rather not " get away" with them lol Id rather have the best machine on the lowest budget...
[11:24:07] <SWPadnos> scales are better than encoders, since they measure actual machine position instead of rotor position
[11:24:25] <GammaX> they use ttl
[11:24:28] <GammaX> I know that.
[11:24:46] <SWPadnos> so if they work correctly, and the backlash is low enough that the motors are controllable using only the scales, then you have the best already
[11:25:22] <GammaX> could backlash make the table take off un expectadly? or would that usualy mean another problem.
[11:25:24] <SWPadnos> to measure backlash, turn on the scales, fire up EMC (making sure that you can see the position feedback), and push really hard on the table
[11:25:27] <SWPadnos> in both directions
[11:25:36] <SWPadnos> if you can see any movement, that's backlash
[11:25:51] <SWPadnos> it's another problem
[11:26:03] <GammaX> tach feedback fromthe servos goes directly to the drive controllers
[11:28:26] <SWPadnos> when does it take off?
[11:28:44] <SWPadnos> (when not supposed to be moving, or when doing moves)
[11:28:55] <GammaX> i put in an incremental value, like .5 it moves a itny bit and then immediatly takes off.
[11:29:16] <SWPadnos> so you tell it to go +0.5, where does it go
[11:29:39] <SWPadnos> +0.5 first, then takes off towards minus, or -0.5 then toward plus, etc. etc
[11:30:08] <GammaX> damn... cant remember I think +.5 and then it takes off +
[11:30:44] <SWPadnos> and maybe if you command -0.5 it goes -0.5ish and then takes off minus?
[11:31:00] <GammaX> correct.
[11:31:14] <GammaX> it happeneds fast.
[11:31:23] <GammaX> almost full speed.
[11:31:54] <SWPadnos> well, looking at the actual motion is important
[11:32:04] <GammaX> have to estop it
[11:32:12] <SWPadnos> no limit switches?
[11:32:32] <GammaX> there are but I hit it before that happeneds
[11:34:07] <SWPadnos> so the table never moves on its own when it's supposed to be stopped?
[11:35:15] <SWPadnos> until you do some of the testing I suggested (using a battery to make a velocity command, etc) it will be difficult to figure out
[11:35:33] <SWPadnos> it could be as simple as a PID coefficient being the wrong sign
[11:36:20] <GammaX> ok. It does move by itself but eeeevver so slowly.
[11:36:32] <GammaX> almost not noticable from the eye
[11:36:55] <SWPadnos> that's different
[11:37:07] <SWPadnos> but if EMC doesn't notice and trip a following error, you have afeedback problem
[11:37:14] <SWPadnos> unless you made FERROR something huge
[11:37:35] <GammaX> i think it did give a following error... I think..
[11:37:57] <SWPadnos> then the amplifiers should have been disabled, if your estop/stop chain are set up right
[11:38:00] <SWPadnos> that's step 1
[11:38:05] <SWPadnos> or step 0 maybe
[11:38:18] <SWPadnos> then you can mess with other stuff with less fear of getting hurt
[11:38:39] <GammaX> well.. they were enabled. I have it setup for them to turn on.
[11:39:01] <SWPadnos> except that a following error should turn them off
[11:39:34] <GammaX> hrmmm I wish i could remember... jease lol
[11:39:55] <SWPadnos> it would be better to discuss it when you're closer to the machine ;)
[11:40:26] <GammaX> Darn.
[11:52:04] <GammaX> SWPadnos, does it sound like anything youve heard before? I was told once that maby the tach signals should be switched..
[11:52:37] <SWPadnos> it does sound like some sort of wire is reversed
[11:52:51] <SWPadnos> but I'd expect it to take off at any time, not just when you command a move
[11:54:25] <GammaX> hmmm
[11:55:00] <SWPadnos> that's one reason to make damned sure that an emc following error will shut off the drives
[11:55:12] <SWPadnos> and that your limit switch setup works
[11:55:47] <SWPadnos> you could then turn on the machine and just leave it. if you come back and it's got a following error, then the problem does happen when it's stopped
[11:56:01] <GammaX> SWPadnos, limit switch is tied in with estop wich is also tied into the power to the drives.
[11:56:17] <SWPadnos> ok
[11:56:18] <GammaX> what would the cause of that most likely be?
[11:56:56] <GammaX> im usin the mesa setup btw
[11:57:09] <SWPadnos> what is "that"?
[11:57:34] <GammaX> whats following error while off?
[11:57:39] <SWPadnos> ah
[11:57:58] <SWPadnos> I didn't say off, I said stopped
[11:58:14] <GammaX> ahh im sorry
[11:58:23] <SWPadnos> so when you turn on the machine, but there is no motion being commanded
[11:58:33] <SWPadnos> it's actually being commanded to stay still
[11:58:38] <SWPadnos> (0 velocity)
[11:58:41] <GammaX> yes
[11:59:08] <SWPadnos> if a following error occurs in that state, then it's not PID tuning that's the (only) problem
[11:59:17] <SWPadnos> probably anyway
[11:59:28] <SWPadnos> are you still overseas?
[12:00:04] <GammaX> yup
[12:00:12] <GammaX> ill be back home sept 9th for 2 weeks
[12:00:34] <SWPadnos> ok. better to discuss it then ;)
[12:00:44] <SWPadnos> if you have time to touch the machine that is
[12:00:53] <GammaX> ok :( haha
[12:01:36] <SWPadnos> it's just that we can't test anything right now, and we don't have all the data we need
[12:01:44] <GammaX> ok
[12:03:58] <GammaX> has there been any more functionality with digitizing yet?
[12:04:24] <SWPadnos> there are some new probe commands, so probably
[12:05:44] <GammaX> I think the tool touchoff would be another great thing to have
[12:09:08] <SWPadnos> there is tool touch-off, but it might not work the way you like
[12:10:17] <GammaX> how so?
[12:10:43] <SWPadnos> I don't know how you want it to work, but there are at least two ways of setting tool lengths
[12:10:51] <SWPadnos> read all about it, I'm not an expert
[12:30:56] <GammaX> SWPadnos, yes u are lol
[12:31:22] <SWPadnos> I'm not an expert, but I think like one ;)
[12:39:25] <GammaX> lol
[12:40:48] <cradek> I play one on TV
[12:41:14] <GammaX> ur on tv?
[12:43:12] <Jymmm> SWPadnos is on TV
[12:43:43] <cradek> GammaX: it's an old joke
[12:44:13] <GammaX> ahhh ok
[13:36:56] <GammaX> anyone know where to get parts for an auto lube pump?
[13:38:50] <SWPadnos> http://www.devcocorp.com/
[13:44:07] <GammaX> thanks
[13:45:02] <Jymmm> http://AutoLubePumpParts.com/
[13:46:14] <Jymmm> Now, if there only was only a http://TwoDollarCNCParts.com
[13:48:05] <Jymmm> But you know shipping would be like $12,333,333
[14:08:10] <GammaX> ok heres another bad question.... what brand lub pump back in the late 80's was yellow? haha
[14:58:44] <GammaX> this is pretty much what my machine is... http://www.cnczone.com/forums/attachment.php?attachmentid=76489&d=1235669706
[14:58:52] <GammaX> any thoughts on what brand that lube pump s?
[15:00:18] <cradek> maybe bijur
[15:00:49] <GammaX> the guy is using mach3 on it wich gecho drives and the same servos
[15:01:07] <cradek> surprising that geckos can push that machine around
[15:02:46] <GammaX> http://www.cnczone.com/forums/showthread.php?t=73820
[15:03:02] <GammaX> cradek theres the 3 page build log if interested. It boggles my mind aswell...
[15:03:32] <cradek> 142v 30amps?
[15:04:23] <cradek> yeah gecko 320 is 80v
[15:04:51] <GammaX> he said he can get 100ipm but i dont think that was only the table, not makin chips
[15:04:53] <cradek> down to 100ipm rapids, he says
[15:05:38] <GammaX> damn.
[15:06:10] <cradek> (to me, that sounds like it just barely works)
[15:06:34] <GammaX> how fast does your machine cut at?
[15:06:56] <cradek> depends what, of course
[15:07:02] <GammaX> alum...
[15:07:04] <cradek> I've cut at .01 ipm and 50 ipm
[15:07:35] <cradek> about 4-20 ipm
[15:07:39] <GammaX> nice
[15:07:45] <GammaX> brb dinner
[15:07:47] <cradek> depends on a lot
[15:08:20] <skunkworks815> cradek: how is jr?
[15:08:37] <cradek> skunkworks815: today I hope to mount stuff
[15:09:10] <cradek> I have fairly full schematics of the magnetics etc. I think it's "just" a matter of wiring it up
[15:09:30] <cradek> I'm slowly figuring out more and more
[15:09:49] <skunkworks815> sweet :)
[15:10:20] <cradek> for the carousel and tool arm motors, they put 60vac into one phase only, as a brake I think
[15:11:28] <cradek> the levels of isolation are amazing - computer to reed relays to ssrs to small relays to big contactors
[15:13:07] <cradek> oddly, the velocity signal was tach -> computer thingy (??) -> amp
[15:13:24] <cradek> I don't know if they had adc to dac, or if it was just wired through
[15:16:38] <skunkworks815> yikes. Do you think they will work with the tack directly into the amps?
[15:16:50] <cradek> I hope so
[15:17:03] <cradek> I don't know what was there - maybe it was just wired through. I don't have that part to examine.
[15:19:25] <cradek> I wish I knew how to make sure I hook up the tach the right direction - that's an easy way to get a (500 ipm) runaway
[15:19:59] <cradek> bbl
[15:34:55] <GammaX> cradek, any idea what model that would be? the auto luber?
[15:42:37] <ilya__> In a lathe mode, X-axis is a length of a detail, and z-axis is a radial direction, isn't it?
[15:57:03] <DaViruz> no, the other way around
[15:57:18] <DaViruz> Z is paralell to the spindle X is perpendicular
[15:57:49] <DaViruz> http://www.dakeng.com/man/TurboCNC_files/image072.jpg
[15:59:44] <ilya__> DaViruz: TurboCNC + FreeDOS = ?
[16:03:50] <ilya__> And `Position 1' of the tool should be applied to the detail by means a negative move by the X-axis?
[16:05:06] <cradek> ilya__: x is radius
[16:05:14] <cradek> or dia
[16:05:37] <cradek> z is out from headstock
[16:06:33] <ilya__> cradek: and position 1 or the 9thof the tool is 'usual'?
[16:11:07] <ilya__> because position 1 seems to mean the `negative' X-value
[16:11:34] <renesis> i used turbocnc and freedos
[16:11:38] <renesis> i gve up
[16:11:48] <renesis> too many weird bugs
[16:12:10] <ilya__> renesis: what is a size of turboCNC package for DOS?
[16:12:25] <ilya__> does it have some GUI?
[16:12:27] <renesis> i dunno, tiny
[16:12:37] <renesis> its terminal gui
[16:12:51] <ilya__> Does it remind the Xemc interface to EMC2?
[16:13:09] <renesis> i dunno ive never used that
[16:13:37] <renesis> ive just used axis and the ugly one
[16:13:42] <ilya__> are there any drawing-to-g code for lathe mode programs?
[16:13:57] <kanzure> inkscape->svg->g-code
[16:14:02] <kanzure> there was an SVG-to-g-code thingy IIRC
[16:14:06] <renesis> mastercam
[16:14:15] <ilya__> anyway, if TurboCNC doesn't work in Windows, I can use free linux EMC2
[16:14:31] <renesis> ive used emc2 for couple years at least
[16:14:54] <renesis> almost everytime its fucked up it turned out to be me doing something stupid
[16:15:00] <ilya__> kanzure: no, for a lathe tool.
[16:15:37] <ilya__> kanzure: there's an SVG-to-g-code thingy in HeeksCNC
[16:15:46] <kanzure> sorry, I don't understand, what is a lathe tool and how is it different from what I suggested?
[16:15:50] <renesis> also windows is horrible for realtime
[16:16:36] <renesis> when i ran mach, my machine would dwell for no reason while windows did shit like blank the screen or generate a tray balloon
[16:17:15] <renesis> meanwhile, on the lunix i can open firefox and browse the web while running a cycle on emc2
[16:18:02] <ilya__> kanzure: for lathe tool, there should be a sequential moves, not just one. Usuallly, 3 mm with 0.2 mm per revolution. Therefore, g-code for lathe tool reminds parallel lines when you create a shaft.
[16:18:35] <renesis> wat
[16:18:37] <kanzure> reminds?
[16:19:09] <ilya__> kanzure: sorry for bad english
[16:19:28] <kanzure> okay, I think I understand
[16:19:29] <kanzure> thanks
[16:19:55] <ilya__> kanzure: run lathe.ini and choose the lathe-pawn.nc in the `examples' directory
[16:20:03] <kanzure> of emc?
[16:20:09] <ilya__> yeah
[16:22:45] <ilya__> I have downloaded some educational books (the ones I used in university) with suggestions for the lathe
[16:23:22] <kanzure> could you recommend some titles? or links by pm?
[16:24:06] <ilya__> kanzure: no, remember I'm Russian? www.listlib.narod.ru
[16:24:58] <ilya__> although it seems illegal to use scanned DJVU books.
[16:26:01] <ilya__> kanzure: i can copy the table and upload to the website
[16:26:08] <kanzure> that would be useful
[16:26:09] <ilya__> *that table
[16:26:14] <ilya__> ok
[16:38:11] <ilya__> uploading thise .tar.gz file to the server ...
[16:48:04] <ilya__> it's only 210 Kb, but it will only be possible to do this later this night
[16:50:12] <ilya__> 2.8 Kb/s, 0 Kb/s, then almost 9 Kb/s... True adventure for those brave ones!
[16:50:59] <ilya__> Mad thing, and connection can be interrupted at any moment
[16:51:20] <BigJohnT> wow you have a worse connection than me
[16:51:35] <ilya__> I'm certainly do.
[16:51:50] <BigJohnT> 49 Kb/s here
[16:51:59] <ilya__> =-O
[17:02:33] <ilya__> Ok, need to restart. Brb.
[17:11:33] <ilya__> kanzure: no, let's here
[17:12:00] <ilya__> the browser updates itself. i will out of money again
[17:18:55] <ilya__> `rough longitudinal': `depth--feedrate--speed in meters per minute for carbon steel with 750 MPa --pig iron with HB 190
[17:19:57] <ilya__> 3--.4|.5|.7|1.--44|38|30|24--29|26|23|20
[17:21:46] <ilya__> It means, to cut 3 mm of carbon steel, we program
[17:26:12] <ilya__> For carbon steel with depth of 3 mm with `G95 F0.4', it is advised to use as high speed, as 44 meters per minute at the place where the cutting bit touches the detail.
[17:26:32] <ilya__> I don't know what does it mean and don't want to.
[17:29:03] <ilya__> Another table is more useful. For diameters ~18 mm for depth <5 mm, 0.25 mm/revolution
[17:31:01] <ilya__> ~30 for <5 mm, .02-.05 mm/r | ~50 for <5 mm, .04-.08 mm/r and for 5-8 mm, .3-.6 mm/r
[17:31:45] <ilya__> this table is for usual steels and pig iron
[17:34:59] <ilya__> Anyway, now there is a question: Can I assign the Spindle speed as a function of current X coordinate?
[17:35:20] <ilya__> Hope I'm online.
[17:39:30] <ilya__> For example, for pig iron with HB 150, for 3 mm depths at .2 mm per revolution, this book offers 92 meters per minute as a speed at wich the material is being cut. So for lathe, in addition to the feedrate, one should normally program the spindle speed as a derivation of the radius or X-coordinate of the end of the tool.
[17:42:55] <ilya__> Can I assign a current X-coordinate to the named parameter?
[17:45:59] <ilya__> Then use the spindle speed = [1000*#<desired_speed>/(that_parameter - #<X_offset_for_the_tool>)] ?
[17:47:05] <ilya__> Will this sophisticated program make a coffee to co-workers?
[18:10:15] <cradek> ilya__: you're talking about constant surface speed (CSS) - see the docs for details
[18:21:39] <geo01005_home> hmm, I wondering if anybody else would be interested in a pyvcp widget that would basically give you access to the canvas widget though the pyvcp xml file ?
[18:22:35] <geo01005_home> (for making custom widgets without modifying the pyvcp_widget.py file directly.
[18:38:43] <roh> geo01005_home which is the canvas?
[18:46:03] <geo01005_home> several of the widgets are derived from the canvas widget like the meter, leds, dial, and others.
[18:47:20] <geo01005_home> http://www.pythonware.com/library/tkinter/introduction/canvas.htm
[18:49:24] <geo01005_home> I'm thinking that for somethings it would be nice to be able to have a canvas widget that you put arcs, lines, ovals, polygons, etc. into.
[18:50:13] <roh> i see.
[18:50:24] <geo01005_home> it would also be nice to have a halpin to switch those object from visible to invisible.
[18:50:39] <roh> well.. i would find it more interresting to have the whole ui as some kind of 'configurable'
[18:50:43] <geo01005_home> Otherwise it is easier to just use .gifs
[18:50:46] <roh> like using glade or so
[18:51:32] <roh> and or more widgets. e.g. vertical bars, or stuff which changes color in not only the meter
[18:53:50] <geo01005_home> yes, it would be nice to make a UI in something like glade. However Glade doesn't give you the option to make new widgets does it?
[18:54:02] <geo01005_home> (I really don
[18:54:09] <geo01005_home> 't know too much about glade)
[18:58:11] <roh> glade allows you to 'config' the whole 'design' and then you just add stuff to the signals and load the config via a small lib on runtime
[19:05:55] <roh> basically it allows you using all gtk widgets and some more
[19:06:25] <geo01005_home> Yeah, so for instance, you couldn't make the current meter widget in glade...(I know the current meter widget is in the Tkinter interface)
[19:06:51] <geo01005_home> You would have to make the widget in code and then get glade to recognize it.
[19:08:51] <roh> japp. but i dont think thats so extremely complicated. i think it would be rather important to find a 'generic' way to encapsule application logic while let the ui be 'editable' with easy tools (the glade editor)
[19:10:44] <SWPadnos> that's not quite as easy as it sounds
[19:11:01] <roh> check http://www.learningpython.com/2006/05/07/creating-a-gui-using-pygtk-and-glade/ for a basic example
[19:11:24] <SWPadnos> the (main) thing that makes an application builder easy to use is when there are controls (buttons, indicators, etc) that already have some function
[19:11:32] <SWPadnos> such as a stop button or a run button
[19:11:43] <roh> one only does the xml file with the ui-tool, naming all signals and so on, generates a python stub, and adds the 'logic' behind it. also the stuf already 'loads' the xml file on runtime then
[19:13:09] <roh> same goes for other languages.. glade can also be used from c etc. doesnt matter much in what language one reacts to a signal
[19:16:11] <roh> in the end it will be 'gtk-builder' .. the technology gets merged into gtk core
[19:16:16] <roh> http://www.micahcarrick.com/05-30-2008/gtk-builder-libglade-faq.html
[19:17:24] <SWPadnos> I think there are roughly three use cases for a UI builder
[19:17:46] <SWPadnos> 1) people who want to add (or subtract) a few things because they don't quite like the way the display looks
[19:18:17] <SWPadnos> 2) people who want to make drastic changes to the UI, perhaps because they want to make a system that is more foolproof to operate (on a production line with a drone running it)
[19:18:48] <SWPadnos> 3) people who want to make new UI elements, probably because they have a need for specific functionality that isn't there yet
[19:19:10] <SWPadnos> pyvcp works well for case 1, not so well for cases 2 and 3
[19:20:02] <SWPadnos> something using gtk/glade/whatever (could be qt also) does well for cases 1 and probably 2, but with #3 being pretty difficult
[19:20:21] <SWPadnos> you have to be a "real programmer" to do 3, probably with any system we could come up with
[19:20:48] <SWPadnos> I think you might have to be a "real programmer" to do #2 as well, with glade or QT
[19:21:11] <SWPadnos> I'm not sure how to make it possible for "normal users" to do #2 or things between #1 and #2
[19:21:38] <roh> SWPadnos case 3 requires you to 'code' the widget you want to use first.. but thats also true now.
[19:21:50] <SWPadnos> yeah, I don't think there's a way around that now
[19:22:10] <SWPadnos> but what's also complex in case 1 and 2 is that there is no inherent function for any UI element with pyvcp
[19:22:10] <geo01005_home> That is why I wanted dirrect access to the canvas widget from the xml file.
[19:22:35] <SWPadnos> all the real work is done in halui (or somewhere), and the connections are what make the UI elements do what you want
[19:22:39] <roh> SWPadnos http://www.pygtk.org/articles/custom-widgets-glade/Custom_PyGTK_Widgets_in_Glade3.html
[19:22:53] <SWPadnos> all of the buttons are just buttons, regardless of whether they're red and say "ESTOP" on them
[19:22:59] <SWPadnos> sure
[19:23:19] <roh> seems easy.. atleast compared to learning tk ;)
[19:23:44] <SWPadnos> there has been some work done on a qt widget set, and I think the EZ-TROL from Smithy uses qt (the source is available somewhere)
[19:25:22] <SWPadnos> there are then the issues of all the bit sof status and control that aren't in HAL
[19:25:31] <SWPadnos> such as position readouts
[19:26:05] <roh> well.. i don't like qt, so he's on his own there
[19:26:22] <geo01005_home> roh only like gtk
[19:26:35] <SWPadnos> and there are interactions between different controls that aren't reflected in the data, such as whether to display absolute or relative positions, or actual vs commanded positions
[19:29:14] <geo01005_home> bbl
[19:30:59] <roh> geo01005_home depends on the usecase. but for complex ui, yes. for custom deep embedded ui with effects required (swoosh shit) enlightment is much better suited, since one can 'define state transitions' there
[19:33:44] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[19:34:13] <Jymmm> Will the REAL SWPadnos please stand up
[19:39:02] <SWPadnos_> SWPadnos_ is now known as SWPadnos