#emc | Logs for 2011-04-03

[00:03:46] -!- theorb [theorb!~theorb@] has joined #emc
[00:04:01] -!- theorbtwo has quit [Ping timeout: 252 seconds]
[00:04:05] theorb is now known as theorbtwo
[00:18:01] -!- sumpfralle [sumpfralle!~lars@on1.informatik.uni-rostock.de] has joined #emc
[00:42:27] -!- jdhNC [jdhNC!~jdh@miranda.org] has joined #emc
[00:43:53] -!- elmo40 [elmo40!~Elmo40@CPE0017ee09ce6d-CM001bd7a89c28.cpe.net.cable.rogers.com] has joined #emc
[00:47:22] -!- skunkworks_ has quit [Remote host closed the connection]
[00:48:20] -!- elmo40 has quit [Ping timeout: 250 seconds]
[00:54:15] -!- roberth__ has quit [Ping timeout: 252 seconds]
[00:55:50] -!- JT-Shop has quit [Quit: ChatZilla [Firefox 3.6.16/20110319135224]]
[01:07:52] -!- sumpfralle has quit [Ping timeout: 264 seconds]
[01:08:06] -!- sumpfralle [sumpfralle!~lars@on1.informatik.uni-rostock.de] has joined #emc
[01:09:05] -!- crazy_imp has quit [Ping timeout: 250 seconds]
[01:11:07] -!- crazy_imp [crazy_imp!~mj@a89-182-205-145.net-htp.de] has joined #emc
[01:40:23] -!- cv has quit [Ping timeout: 240 seconds]
[01:58:40] -!- Valen has quit [Quit: Leaving.]
[02:11:10] <jdhNC> I turned down a quarry dive for tomorrow
[02:11:16] <jdhNC> <urk>
[02:31:42] -!- jdhNC [jdhNC!~jdh@miranda.org] has parted #emc
[02:54:49] -!- sumpfralle has quit [Ping timeout: 250 seconds]
[03:12:50] -!- tris has quit [Ping timeout: 260 seconds]
[03:18:44] -!- tris [tris!~tristan@173-164-188-122-SFBA.hfc.comcastbusiness.net] has joined #emc
[03:54:01] -!- psha [psha!~psha@] has joined #emc
[03:55:02] -!- EDocTooR has quit [Quit: Leaving]
[04:27:25] -!- ve7it has quit [Remote host closed the connection]
[04:58:36] -!- psha has quit [Quit: Lost terminal]
[05:07:17] -!- mhaberler [mhaberler!~mhaberler@extern-182.stiwoll.mah.priv.at] has joined #emc
[05:31:59] -!- psha [psha!~psha@] has joined #emc
[06:41:44] -!- psha has quit [Read error: Connection reset by peer]
[06:48:31] -!- Petsku [Petsku!~Petteri@a88-112-195-37.elisa-laajakaista.fi] has joined #emc
[06:54:48] -!- awallin_ has quit [Remote host closed the connection]
[06:57:45] -!- Petsku has quit [Quit: Petsku]
[07:00:47] -!- Connor [Connor!~Connor@] has parted #emc
[07:00:56] -!- Connor [Connor!~Connor@] has joined #emc
[07:37:32] -!- EDocTooR [EDocTooR!~EDocTooR@206-248-162-24.dsl.teksavvy.com] has joined #emc
[07:51:40] -!- robh__ [robh__!~robert@5ace70bc.bb.sky.com] has joined #emc
[07:55:55] -!- EDocTooR has quit [Read error: Connection reset by peer]
[08:00:47] -!- EDocTooR [EDocTooR!~EDocTooR@206-248-162-24.dsl.teksavvy.com] has joined #emc
[08:04:55] -!- EDocTooR has quit [Read error: Connection reset by peer]
[08:19:26] -!- cncbasher [cncbasher!~kvirc@cpc8-hart9-2-0-cust686.11-3.cable.virginmedia.com] has joined #emc
[08:22:29] -!- Valen [Valen!~Valen@ppp121-44-184-217.lns20.syd7.internode.on.net] has joined #emc
[08:35:18] -!- cncbasher has quit [Ping timeout: 264 seconds]
[08:48:38] -!- cncbasher [cncbasher!~kvirc@cpc8-hart9-2-0-cust686.11-3.cable.virginmedia.com] has joined #emc
[08:55:55] -!- EDocTooR [EDocTooR!~EDocTooR@206-248-162-24.dsl.teksavvy.com] has joined #emc
[09:00:02] -!- cncbasher [cncbasher!~kvirc@cpc8-hart9-2-0-cust686.11-3.cable.virginmedia.com] has parted #emc
[09:07:57] -!- cncbasher [cncbasher!~kvirc@cpc8-hart9-2-0-cust686.11-3.cable.virginmedia.com] has joined #emc
[09:23:25] <fragalot> :'( dug out my old diy controller just now
[09:23:35] <fragalot> apparently the X and Z axis controllers have given up
[09:25:09] -!- e3m [e3m!~IceChat7@] has joined #emc
[09:27:56] <fragalot> scrap that, X is dead, Y and Z appear to work
[09:28:05] * fragalot decides to take the whole thing apart
[09:30:12] -!- nullie [nullie!~nullie@nullie.telenet.ru] has joined #emc
[09:40:26] -!- isssy [isssy!~isssy@78-83-51-185.spectrumnet.bg] has joined #emc
[10:03:55] -!- e3m has quit [Quit: Clap on! , Clap off! Clap@#&$NO CARRIER]
[10:08:30] -!- TekniQue [TekniQue!baldur@unaffiliated/teknique] has joined #emc
[10:13:47] <fragalot> hurray fixed it
[10:40:16] <fragalot> Thought it operates at a "Close enough" level :P (current limitation isn't what you'd call synchronous) one coil will get 200mA while the other gets 180 >.>
[10:55:01] <crazy_imp> heyho
[10:56:04] <crazy_imp> we've discovered some problems emc and this http://amalas.eu/~slime/PROJECTS/CNC/heimatstube/test.ngc gcode
[10:56:55] <crazy_imp> emc displays the correct form, but it drives of to do a huge circle...
[10:57:16] -!- slime [slime!~slime@dslb-088-070-242-079.pools.arcor-ip.net] has joined #emc
[10:57:21] <crazy_imp> at line 15 the madness begins...
[10:57:22] <slime> re
[10:57:23] slime is now known as Guest5824
[10:57:30] <crazy_imp> hey slime :D
[10:57:34] <Guest5824> :D
[10:59:00] <Guest5824> http://amalas.eu/~slime/PROJECTS/CNC/heimatstube/P1010082.JPG
[10:59:10] <Guest5824> http://amalas.eu/~slime/PROJECTS/CNC/heimatstube/P1010084.JPG
[10:59:30] -!- elmo40 [elmo40!~Elmo40@CPE0017ee09ce6d-CM001bd7a89c28.cpe.net.cable.rogers.com] has joined #emc
[10:59:32] <archivist> crazy_imp, there was a mention of an arc problem in the last few weeks and I think Ive seen a fix done
[11:00:00] <crazy_imp> archivist: just tested it with emc 2.6 / git version and it's the same
[11:02:25] <crazy_imp> archivist: http://hannover.ccc.de/~imp/emc_drifting.png
[11:05:46] <archivist> looks a good test case to attach to a bug report
[11:06:18] <crazy_imp> i'm search right now if there's already an open bug
[11:07:36] <crazy_imp> maybe http://sourceforge.net/tracker/?func=detail&aid=1922330&group_id=6744&atid=106744 ?
[11:17:56] <lilalinux> maybe just too high acceleration?
[11:18:59] <crazy_imp> ?
[11:22:24] <Guest5824> oO
[11:23:42] Guest5824 is now known as slime2k
[11:52:23] * jthornton wonders how can you have a GPL Version 2 or later and a Copyright on a bit of code
[11:54:05] <archivist> gpl is a licence you own copyright if you create
[11:54:59] <archivist> crazy_imp, yes probably , I must have spotted the comment on the list and it stuck in the rear of the grey matter
[11:57:23] <crazy_imp> archivist: user or dev ml?
[11:58:06] <archivist> crazy_imp, see date of cradek's comment on that bug, thats what I would have seen
[11:58:33] <archivist> crazy_imp, looking at your last pic are you sure its axis and not cambam
[12:02:48] <crazy_imp> archivist: it's slime2k pic ;). from my point of view, if it's cambams fault, emc should render it inside the toolpath as motion of the toolhead, right? but emc doesn't display this movement until it performs the movement
[12:04:19] <crazy_imp> (either way, emc should display it OR dont do it ;)
[12:04:44] -!- newbynobi [newbynobi!~norbert@p4FC9CA80.dip.t-dialin.net] has joined #emc
[12:13:03] <newbynobi> mhaberler: I am trying to integrate an own gladevcp panel as an tab into my axis configuration. Whal ever I tried result in one error line "invalid Tap configuration" To be sure that the error dos not come from my ui. file I tried also with the test panel. At this time I tried the following: I copied the gladevcp-test.hal , gladevcp-test.glade and gladevcp-test.ui into my config dir and addes the following lines to my ini file.EMBED_TAB = Glad
[12:13:04] <newbynobi> eVCP
[12:13:04] <newbynobi> EMBED_TAB_COMMAND = gladevcp -c gladevcp-test -H gladevcp-test.hal -w {XID} gladevcp-test.glade and as written in the documentation a comented out the loadusr lines.
[12:13:15] <newbynobi> can somebody help?
[12:15:01] <newbynobi> If I just write GLADEVCP=ToolChangeTest.ui my Panel is loaded left side of axis.
[12:17:32] <anonimasu> crazy_imp: agreed, displayed motion should equal motion out
[12:18:57] <archivist> crazy_imp, bug is in the version of emc I have on this box too, just ran the test case on axis_mm it draws a full circle
[12:25:45] -!- Valen has quit [Quit: Leaving.]
[12:26:04] <crazy_imp> errr
[12:26:13] <crazy_imp> i can't attach files?
[12:30:21] -!- sumpfralle [sumpfralle!~lars@on1.informatik.uni-rostock.de] has joined #emc
[12:33:33] -!- Farthen has quit [Read error: Operation timed out]
[12:42:31] <archivist> http://www.collection.archivist.info/Screenshot-29.png why was there no error when it went out of the box too
[12:48:34] -!- sumpfralle has quit [Read error: No route to host]
[12:49:28] -!- sumpfralle [sumpfralle!~lars@on1.informatik.uni-rostock.de] has joined #emc
[12:49:46] -!- Farthen [Farthen!~Farthen@static.] has joined #emc
[12:51:46] -!- sumpfralle has quit [Client Quit]
[12:53:19] <crazy_imp> archivist: hm, here it stops with Exceeded positive soft limit on joint 1
[13:00:05] <archivist> it was running un homed and not touched off, it does error now
[13:00:52] <archivist> http://www.collection.archivist.info/Screenshot-30.png
[13:03:33] -!- mhaberler_ [mhaberler_!~mhaberler@imac.stiwoll.mah.priv.at] has joined #emc
[13:15:03] -!- nullie has quit [Ping timeout: 240 seconds]
[13:23:07] -!- andrus [andrus!~quassel@] has joined #emc
[13:26:55] -!- andrus has quit [Remote host closed the connection]
[13:33:50] -!- nullie [nullie!~nullie@nullie.telenet.ru] has joined #emc
[13:40:45] -!- mhaberler_ has quit [Quit: mhaberler_]
[13:50:23] <slime2k> but
[13:50:40] <slime2k> the question is why does it draw the arc?
[14:12:14] -!- sumpfralle [sumpfralle!~lars@on1.informatik.uni-rostock.de] has joined #emc
[14:13:12] <archivist> slime2k, bug in the trajectory planner maths by the look of it
[14:17:12] <slime2k> ?
[14:26:46] EDocTooR is now known as EDGuest
[14:28:59] -!- EDGuest has quit [Quit: Leaving]
[14:30:22] <archivist> slime2k, odd comments in the source like // absolute value of big_radius, I wonder why thats there
[14:31:50] -!- EDocTooR [EDocTooR!~EDocTooR@206-248-162-24.dsl.teksavvy.com] has joined #emc
[14:41:24] -!- tlab [tlab!~tlab@c-69-243-187-136.hsd1.in.comcast.net] has joined #emc
[14:42:35] -!- tlab has quit [Client Quit]
[14:44:38] <archivist> I would want to run it under a debugger to watch the variables, seems to be making a bad choice circle v semicircle
[15:03:07] <anonimasu> hello
[15:03:20] <anonimasu> anyhone got time to help me out a bit
[15:03:38] <anonimasu> I have a 7i37com hooked up to the second port onm y 7i47
[15:03:45] <anonimasu> err my 7i47
[15:04:04] <anonimasu> and im wondering how to set it to give me some output in some pin(manually for testing)
[15:05:54] -!- acemi [acemi!~acemi@unaffiliated/acemi] has joined #emc
[15:06:42] <cradek> the arc on line 15 is .001005 mm long with radius of 500ish mm
[15:07:12] <slime2k> is this a problem?
[15:07:22] <cradek> when the endpoints are very close together, emc guesses whether the programmer meant tiny arc or full circle. unfortunately it guesses wrong in this case
[15:08:32] <cradek> (actually AXIS and motion guess differently, which may be the actual bug we can fix, IMO)
[15:09:09] <cradek> can you ask cambam not to generate these tiny moves?
[15:09:15] <anonimasu> hm.. everything is wired but i have no clue next
[15:09:29] <anonimasu> how can I get power out on a pin on p3?
[15:09:58] <slime2k> cradek: i think no :D
[15:13:52] <cradek> http://timeguy.com/cradek-files/emc/axis-slime2k.png
[15:14:05] * anonimasu sighs
[15:14:14] <cradek> here's AXIS drawing the code around lines 15-17
[15:14:28] <cradek> the gcode is very weird there
[15:14:30] <anonimasu> how do i set a specific pin on the p3? do i need to find what it is in the manual of the board(7i47)
[15:15:12] <cradek> bbl
[15:15:14] <cradek> breakfast
[15:16:05] <slime2k> in cambam i found an option called "ArcFitTolerance"
[15:16:32] <slime2k> "Maximum allowed error when arc fitting"
[15:16:46] <slime2k> but i dont know what it mean
[15:17:36] <pcw_home> Then 7I43 manual will tell you the GPIO # to physical pin association (as will the .PIN file)
[15:17:37] <pcw_home> the GPIO pins that are used as outputs need to be set to output mode (man hostmot2)
[15:18:00] <anonimasu> should I have 5v encoder power just by hooking boards togther?
[15:18:06] <anonimasu> or do I need to turn that output on?
[15:18:17] <pcw_home> Yes is the 7I43 is powered
[15:18:25] <pcw_home> if
[15:18:28] <anonimasu> then my board seems doa :S
[15:18:45] <anonimasu> the 7i37-com
[15:18:48] <anonimasu> no leds or anything
[15:19:22] <pcw_home> there are no LED on the 7I37COM
[15:19:37] <anonimasu> it looked like one in the corner
[15:19:48] <anonimasu> and I should have the encoder power from it when its plugged?
[15:19:53] <anonimasu> or is that only if I enable the pin?
[15:20:18] <pcw_home> Encoder power?
[15:20:26] <anonimasu> the 5v output
[15:20:30] <anonimasu> -_-
[15:20:33] <anonimasu> wrong board!
[15:20:52] <anonimasu> im talking about the 7i47 one
[15:23:00] <pcw_home> OK the 7I47 has some encoder LEDs . Is the 7I47 jumpered for cable power?
[15:24:29] <pcw_home> If you have not moved any jumpers and the FPGA card is powered, you should have power to the 7I47 (this is the default)
[15:24:39] <pcw_home> bbl
[15:25:05] <anonimasu> err it works now
[15:25:07] <anonimasu> sorry.
[15:25:22] <anonimasu> next issue how do i set power out on the other board the 7i37com
[15:27:11] <pcw_home> You connect the V+ terminal block to your field power (5V,12V,24V) and set a GPIO output pin low to turn on the output)
[15:27:22] <pcw_home> really bbl, being paged...
[15:27:43] <anonimasu> allright
[15:27:49] <anonimasu> from wha tdo i set it?
[15:29:54] <bzzzz> yay, the git repo is back up
[15:30:06] -!- psha [psha!~psha@] has joined #emc
[15:30:46] <bzzzz> who hosts it, anyway?
[15:34:00] <pcw_home> anonimasu: make sure you connect field power to TB4, an do P3 the GPIO that controls the 8 outputs is IO40 through IO47
[15:34:22] <pcw_home> s/an do/and on/
[15:37:18] <anonimasu> how can i set that manually?
[15:37:24] <anonimasu> halcmd setp?
[15:45:14] <anonimasu> now the qustion is wher edo i start with the thc...
[15:45:44] <anonimasu> it's wired like f+ f- and g to the other board the 7i37
[15:45:51] <anonimasu> how can i monitor the signal?
[15:46:33] -!- mhaberler has quit [Quit: mhaberler]
[15:49:50] -!- kb8wmc [kb8wmc!~chatzilla@] has joined #emc
[15:50:14] -!- mhaberler [mhaberler!~mhaberler@imac.stiwoll.mah.priv.at] has joined #emc
[15:50:45] -!- mhaberler_ [mhaberler_!~mhaberler@macbook.stiwoll.mah.priv.at] has joined #emc
[15:55:17] -!- isssy has quit [Read error: Connection reset by peer]
[15:57:37] <anonimasu> i cant see input from my thc board
[16:01:28] -!- isssy [isssy!~isssy@78-83-51-185.spectrumnet.bg] has joined #emc
[16:07:00] -!- isssy has quit [Quit: Visitor from www.linuxcnc.org]
[16:08:35] -!- isssy [isssy!~isssy@78-83-51-185.spectrumnet.bg] has joined #emc
[16:13:57] <crazy_imp> cradek: even if emc guesses the wrong option (small arc vs. full circle), it should display the path - are there two different gcode interpreting instances?
[16:14:50] <anonimasu> brbhm..
[16:14:56] <anonimasu> it works but no input
[16:23:19] -!- JT-Shop [JT-Shop!~chatzilla@216-41-156-49.semo.net] has joined #emc
[16:30:37] <cradek> slime2k: you can fix that gcode by just removing the lines 15,16,17 which make the bogus little shape shown in my screenshot
[16:33:29] <cradek> crazy_imp: yes I agree they should match.
[16:36:08] <anonimasu> hm.. everything works except for the thc input which I cant find
[16:36:32] <JT-Shop> did you load the thc comp?
[16:36:43] <anonimasu> I cant see the input working at all
[16:36:55] <anonimasu> im htinking my 7i44 needs the termination jumper moved or something
[16:37:11] <anonimasu> since both pin 24/25 was true
[16:37:13] <anonimasu> constantly
[16:38:06] <JT-Shop> are you looking at the output from the TCH card with halscope or a real scope?
[16:38:13] <anonimasu> halscope
[16:38:18] <anonimasu> the output from the card is there
[16:42:43] -!- IchGuckLive [IchGuckLive!~chatzilla@95-89-104-26-dynip.superkabel.de] has joined #emc
[17:01:32] -!- kb8wmc has quit [Ping timeout: 255 seconds]
[17:01:52] cjdavis1 is now known as cjdavis
[17:02:35] -!- kb8wmc [kb8wmc!~chatzilla@] has joined #emc
[17:04:21] -!- isssy has quit [Quit: Visitor from www.linuxcnc.org]
[17:06:58] <slime2k> cradek: hmm, i cant check every g-code i made
[17:07:19] -!- andypugh [andypugh!~andy2@cpc2-basl1-0-0-cust1037.basl.cable.virginmedia.com] has joined #emc
[17:07:36] <slime2k> thats a workaround, but no solution :D
[17:11:10] -!- i_tarzan has quit [Ping timeout: 246 seconds]
[17:11:29] -!- i_tarzan [i_tarzan!~i_tarzan@] has joined #emc
[17:14:43] -!- ve7it [ve7it!~LawrenceG@S0106009027972e37.pk.shawcable.net] has joined #emc
[17:15:16] <slime2k> ok, found the solution
[17:15:19] <slime2k> @cambam
[17:15:35] <slime2k> u could define the minimum arc length
[17:15:58] <slime2k> changed it from 0.0001 to 0.01
[17:16:00] <crazy_imp> slime2k: but even then, it's just a workaround :P
[17:20:36] <TekniQue> heh, so the bug is that it's trying to cut a full circle
[17:20:50] <TekniQue> when the arc becomes so short that it gets rounded off in the math
[17:20:52] <TekniQue> to zero
[17:21:10] <TekniQue> nice
[17:21:45] <crazy_imp> TekniQue: maybe. anyway you can also pinpoint the bug to: axis doesn't show the circle before it's too late.
[17:21:53] <IchGuckLive> therfore most CAM systems have a button for 4 arcs to the full circel
[17:22:47] <IchGuckLive> or a other post to define this
[17:22:56] <TekniQue> crazy_imp: yeah, my gut feeling is that the code that controls the machine is not the same code as draws the preview
[17:23:04] <TekniQue> without having looked at the code
[17:23:27] <TekniQue> or that the rounding off happens when the path is converted to some units in the machine control code
[17:23:41] <IchGuckLive> on my simple pythons this is the case there are dividet
[17:23:41] <crazy_imp> TekniQue: or the preview rendering part just ignores those stuff because it's to small for preview, dunno
[17:24:44] -!- skunkworks [skunkworks!~chatzilla@str-bb-cable-south2-static-6-412.dsl.airstreamcomm.net] has joined #emc
[17:27:39] <IchGuckLive> *INFO* Heekscad has now 4 more Postprozessor available from the Saturday compiling !
[17:27:42] <andypugh> An M2 or M3 with no movement in X or Y is a full circle. I assume that you might be falling foul of two interpreters which disagree on how small. (Or did you already say that?)
[17:28:20] <IchGuckLive> G2/G3 ?
[17:29:00] <IchGuckLive> M3 starts the tool
[17:29:54] <IchGuckLive> you need at least a IJ or R
[17:31:09] <TekniQue> G2 F2500.0 X615.8703 Y184.685 I-588.0936 J514.337
[17:31:09] <TekniQue> G2 X615.8702 Y184.6849 I-578.7604 J524.8171
[17:31:19] <TekniQue> yeah here's your problem
[17:31:40] <TekniQue> a movment of 0.0001 in X and 0.0001 in Y
[17:32:01] <TekniQue> which likely gets rounded off below 1 step
[17:32:04] <TekniQue> if it's a stepper machine
[17:32:19] <IchGuckLive> *info* this mashines are now available -> http://code.google.com/p/heekscnc/source/browse/trunk/nc/machines.txt?spec=svn1315&r=1315
[17:33:19] <TekniQue> my own stepper machine only has a resolution of 0.003mm for example
[17:33:30] <TekniQue> with 5mm/rotation pitch ball screws
[17:33:36] <TekniQue> and 8x microstepping
[17:35:20] <IchGuckLive> is this in absolut G90 ? and where is the Income sdentence ? also CRC in use ?
[17:36:32] <andypugh> Yeah, sorry, I meant G2 and G3. The interpreter says that start X and end X are identical, and so performs a full circle.
[17:37:12] <andypugh> There needs to be _some_ tolerance, because floating point numbers are a bit wooly where equality is concerned.
[17:37:19] <TekniQue> yes
[17:37:24] <anonimasu> you cant quite tell cam programs to limit the output
[17:37:43] <anonimasu> since you dont have any clue as to whatever the machine your working on is working at
[17:37:55] <anonimasu> _dont_ do them is really just a hack...
[17:38:22] <TekniQue> the real problem there is that the preview and the machine control don't agree
[17:38:26] <anonimasu> yes
[17:38:27] <TekniQue> on how to interpret the code
[17:38:28] <andypugh> There should be a minumum segment length setting in any rational CAM system.
[17:39:32] <anonimasu> well, on my heidenhain i can do 0.001-0.003mm circles
[17:39:42] <anonimasu> it dosent complain or do weird things...
[17:39:47] <andypugh> A 10um arc on a 615mm component makes little sense. (it might as well be a straight line)
[17:39:59] <anonimasu> so i expect emc to handle it better :)
[17:40:13] -!- nullie has quit [Quit: Ex-Chat]
[17:40:21] <andypugh> Yes, but can you do .003 long arcs with a 600mm radius?
[17:41:00] <anonimasu> probably...
[17:41:09] <anonimasu> atleast the control wont screw up
[17:41:10] <anonimasu> :)
[17:41:21] <anonimasu> it will interpolate down to whatever resolution i have
[17:41:29] -!- mhaberler has quit [Quit: mhaberler]
[17:41:29] mhaberler_ is now known as mhaberler
[17:41:41] -!- mhaberler_ [mhaberler_!~mhaberler@imac.stiwoll.mah.priv.at] has joined #emc
[17:42:13] <andypugh> Yes, but, the converse problem exists too, there is a risk of full-circle arcs being interpreted as vanishingly small moves.
[17:42:38] <TekniQue> exactly
[17:42:41] <anonimasu> hm, im 99.9% sure my control would give me a proper arc
[17:42:47] <anonimasu> add another 9
[17:43:10] <cradek> any control that uses floating point will have numerical instability at some point
[17:43:37] <andypugh> well, here we are looking at a 99.99999999% problem...
[17:43:40] <cradek> it would be better to have different gcodes for full circles, but we're stuck with g2/g3
[17:44:33] <andypugh> I agree that the preview and the motion planner should agree on how small is small, though.
[17:44:41] <cradek> g2/g3 with endpoints VERY close can mean either a tiny or a huge arc
[17:44:52] <IchGuckLive> andypugh: agree in 99% it works fine
[17:44:53] <cradek> yes I do too
[17:47:28] -!- IchGuckLive [IchGuckLive!~chatzilla@95-89-104-26-dynip.superkabel.de] has parted #emc
[17:48:10] -!- mhaberler_ has quit [Quit: mhaberler_]
[17:48:15] -!- mhaberler has quit [Quit: mhaberler]
[17:48:27] -!- skunkworks has quit [Ping timeout: 276 seconds]
[17:49:43] <crazy_imp> think we are drifting away from the problem of incorrect displaying ;)
[17:50:31] <andypugh> I was trying to edge back to it.
[17:52:06] <crazy_imp> :)
[18:01:18] <cradek> if (circle->angle > -(CIRCLE_FUZZ) && circle->angle < (CIRCLE_FUZZ)) {
[18:01:18] <cradek> circle->angle = PM_2_PI;
[18:01:18] <cradek> }
[18:01:36] <cradek> ...
[18:01:36] <cradek> ...#define CIRCLE_FUZZ (0.000001)
[18:01:36] <cradek> /* Bug fix for the missing circles problem */
[18:02:27] <cradek> so if the arc is less than .000001 radians, it gets changed into a full circle
[18:03:25] <TekniQue> yup
[18:03:29] <cradek> and what someone said earlier appears to be true: it was done to fix a problem of missing full circles that someone expected to see.
[18:03:50] <andypugh> Which seems reasonable. (or, rather, commanding an arc so small seems unreasobnable)
[18:04:00] <cradek> can't win, can we
[18:04:04] <psha> cradek: which lead to opposite issue
[18:04:25] <psha> if you get very small arc with very large radius you'll end with Huge circle :)
[18:04:32] <andypugh> So, the question is, why isn't that shown in preview?
[18:05:03] <cradek> because AXIS doesn't have identical fuzz
[18:06:09] <andypugh> Seems like an easy fix, then we have bug-free behaviour, which is documented, and it's the users problem if they expected something different :-)
[18:06:15] <cradek> if you look closely at line 16 you can see it goes back in the opposite direction for a tiny bit
[18:06:22] <cradek> (16 of the gcode)
[18:06:32] <crazy_imp> i fear some day some other problems will appear with the circle_fuzz -> http://nopaste.info/2a8a77c1e2.html multiple definitions...
[18:06:34] <cradek> andypugh: it's anything but easy to make them agree in all cases of numeric instability
[18:07:06] <andypugh> It is, without a doubt, a a wrong behaviour on the part of the CAM program.
[18:07:24] <cradek> yeah I think we all agree the gcode is crap
[18:07:37] <cradek> (but that's kind of beside the point)
[18:07:38] <TekniQue> cambam?
[18:08:27] <cradek> if that's what generated it, yes
[18:08:27] <cradek> crazy I don't understand what you're saying
[18:08:30] <cradek> crazy_imp: ^
[18:09:34] <crazy_imp> cradek: take a look at the paste. there are two different files which define CIRCLE_FUZZ (0.000001) - some day somebody may change one value and miss the other -> new problems...
[18:09:44] <cradek> those are the same file
[18:10:01] <TekniQue> is one actually a symlink to the other?
[18:10:10] <cradek> /include has copies
[18:10:15] <TekniQue> ok
[18:10:45] <crazy_imp> ack :)
[18:10:57] <andypugh> Can somebody revert 0b113e98a1621d3e8aa6467eb535b514f0da3d5b
[18:11:00] <andypugh> ?
[18:11:48] <cradek> on what branch?
[18:11:56] <andypugh> Dunno
[18:12:29] <andypugh> I think it only affected the master of the time, so 2.5 and 2.6 now I guess
[18:13:05] <andypugh> I don't want the wrong-way-scale gearchange escaping into a production release
[18:15:26] <cradek> on 2.5 branch, looks like
[18:16:00] <cradek> working on it
[18:18:34] -!- rooks has quit [Ping timeout: 246 seconds]
[18:18:45] <cradek> if I ruled the world, I might make it so you can only have a full circle if you do NOT respecify the endpoint (in G17, that would mean no X or Y words, only IJ)
[18:19:13] <cradek> of course that will break CAM programs
[18:21:41] <TekniQue> agreed
[18:22:01] <TekniQue> I'd say this is a design flaw of the G-code language
[18:22:34] <TekniQue> to really avoid it you must split your full circles into two or more arcs
[18:23:42] <TekniQue> and of course, not try to make arcs with such a tiny angle
[18:23:58] <TekniQue> the first one is easy
[18:24:04] <TekniQue> the latter, not as easy
[18:24:40] <andypugh> I suspect that the first G-code software/hardware used fixed-point BCD so there was no question of rounding errors.
[18:24:50] <TekniQue> because it means some kind of arbitrary limit implemented in the CAM
[18:25:20] <andypugh> It is when you start introducing calculations which can return almost-but-not-quite zero results that the trouble starts.
[18:25:43] <TekniQue> yes, typical floating point problem
[18:25:45] <cradek> maybe someone smarter about FP than me can fix pmCircleInit to be more stable in this situation
[18:26:04] <andypugh> Does such a person exist?
[18:26:13] <cradek> hell yes :-)
[18:27:00] <andypugh> Yeah, OK, so the guys who wrote IEEE 754 count. I meant in terms of EMC2 devs.
[18:31:19] <cradek> looks like AXIS uses 1e-12 in a way that motion uses CIRCLE_FUZZ (1e-6)
[18:33:26] <cradek> yes I get a full circle in the preview if I make them match.
[18:33:30] -!- rooks [rooks!~rooks@102-bem-18.acn.waw.pl] has joined #emc
[18:33:36] <cradek> I'm tempted to call that a fix
[18:34:52] <crazy_imp> .oO(and bad architecture if you have to keep an eye on your constants so they match ;)
[18:35:35] <cradek> .oO(he's never worked on a large and old code base)
[18:36:08] <TekniQue> .oO(But EMC isn't a large and old code base by my standards)
[18:37:07] <andypugh> Some segments of the code use eps
[18:37:09] <crazy_imp> cradek: i did... with great pain, but usually the biggest strugle is to get the crap compiling
[18:46:16] <cradek> http://git.linuxcnc.org/gitweb?p=emc2.git;a=commitdiff;h=6cbafb54d7e1e0437bdfbdb0f56c04c77f587805
[18:46:53] -!- toastydeath [toastydeath!~toast@c-69-140-223-139.hsd1.md.comcast.net] has joined #emc
[18:48:55] <archivist> cradek, should that go in all :) and be documented what the fuzz value happens to be :)
[18:49:15] <archivist> in the g3,3 docs
[18:49:23] <archivist> g2,g3
[18:49:41] -!- distracted [distracted!~distracte@] has joined #emc
[18:49:49] <cradek> nobody's going to manually check their CAM output for arcs traversing less than 1e-6 radians
[18:50:26] <archivist> no but they can see a circle and read the docs
[18:50:34] <TekniQue> as long as the preview is consistent with the machine control
[18:50:42] <TekniQue> it's a non-issue IMO
[18:50:51] <archivist> hence backport to older versiions too
[18:50:52] <TekniQue> (thank god we have a preview)
[18:51:28] <TekniQue> I'm always slightly nervous working on machines without a preview
[18:52:16] <archivist> preview is still wrong in older versions
[18:52:34] -!- Birdman3131 has quit [Read error: Connection reset by peer]
[18:52:35] <cradek> I can't say this change will make them always match.
[18:53:16] <archivist> I can see that with float issues
[18:53:45] <cradek> the two chunks of code are completely different
[18:54:49] <cradek> bbl, out of battery.
[18:58:34] <fragalot> I don't think any of the CAM tools I use output arcs anyway =P
[18:58:38] <fragalot> just G0 and G1's
[18:58:54] <fragalot> (yes the ngc files get big)
[18:59:54] <TekniQue> it's things like this that make me avoid floting point wherever I can
[19:01:41] <crazy_imp> hehe
[19:03:37] <crazy_imp> thanks for troubleshooting :)
[19:04:00] <TekniQue> but I'm a microcontroller, FPGA and hardware driver guy
[19:04:11] <TekniQue> so I rarely need FP
[19:14:13] <fragalot> can't really get around 'm when doing things like FFT's though
[19:15:05] <andypugh> I think I have seen an all-integer FFT.
[19:17:33] <fragalot> last DSP I worked with was all-floats :P
[19:17:45] <fragalot> couldn't use values > 1
[19:18:03] <fragalot> well, could only use < 1
[19:31:34] <anonimasu> JT-Shop: are you there?
[19:33:30] <jthornton> I'm down here at the moment
[19:34:25] <jthornton> but heading back to the shop as soon as the printer is finished
[19:34:39] <anonimasu> how did you get the signals for the thc into your computer?
[19:35:06] <anonimasu> (the diff inputs on the mesa 7i33) ?
[19:35:15] <jthornton> the 5i20
[19:35:28] <anonimasu> err 7i37 i mean
[19:35:38] <anonimasu> err damn..
[19:35:45] <anonimasu> 7i34.. rhe rs422 interface..
[19:36:21] <jthornton> I connected the THC card to the 5i20 as I recall
[19:36:28] <anonimasu> no input board at all?
[19:36:39] <jthornton> I can peek in there in a bit
[19:36:50] <jthornton> I made a terminal board
[19:37:33] <jthornton> http://i47.photobucket.com/albums/f163/johnplctech/50pin01.jpg
[19:37:36] <anonimasu> Ihttp://i47.photobucket.com/albums/f163/johnplctech/50pin01.jpg
[19:37:37] <anonimasu> http://i47.photobucket.com/albums/f163/johnplctech/50pin01.jpg
[19:37:38] <anonimasu> whoops
[19:37:44] <anonimasu> i have the mesa board with terminal strips
[19:38:10] <jthornton> you have a 7i37TA?
[19:38:18] <anonimasu> 7i37com
[19:38:23] <anonimasu> for this machine
[19:38:33] <anonimasu> but im not using that for the input for the plasma
[19:38:38] <anonimasu> im using the 7i34 for that
[19:38:41] <jthornton> I'm not familiar with that one
[19:38:44] <anonimasu> but I cant see anything on the differential inputs
[19:38:55] <anonimasu> maybe im reading them wrong or something
[19:39:16] <jthornton> your trying to see the frequency output from the THC card?
[19:39:21] <anonimasu> yes i am
[19:39:25] <Jymmm> jthornton: YEAH BABY... THC!!!
[19:39:42] <anonimasu> in halscope
[19:40:29] <jthornton> as I recall that signal went to an encoder input (after some scaling) and the velocity output from the encoder is what I sent to the thc comp
[19:42:27] <anonimasu> i cant figure out where to check for it in halscope
[19:42:42] <anonimasu> it's supposed to be pin 25 on the card but all i see is a constant high level
[19:42:46] <anonimasu> for 25/26/27 and so on
[19:43:48] <jthornton> I'm not that good with halscope, if you shut off the power to the THC card does it go to 0?
[19:45:10] <anonimasu> no
[19:46:30] <jthornton> that does not sound correct
[19:48:09] <anonimasu> they are differential inputs that's why im wondering if someone did the same
[19:49:42] -!- psha has quit [Quit: leaving]
[19:53:02] -!- mhaberler [mhaberler!~mhaberler@imac.stiwoll.mah.priv.at] has joined #emc
[19:53:34] <anonimasu> figured it out, i need termination for my receiver
[19:53:58] -!- mhaberler_ [mhaberler_!~mhaberler@macbook.stiwoll.mah.priv.at] has joined #emc
[19:54:14] <jthornton> cool
[20:17:38] -!- pjm__ [pjm__!~pjm@host81-146-71-253.btremoteinternet-dsl.bt.net] has joined #emc
[20:20:43] -!- pjm has quit [Ping timeout: 240 seconds]
[20:23:39] -!- ries [ries!~ries@] has joined #emc
[20:28:43] -!- JT-Shop has quit [Quit: ChatZilla [Firefox 3.6.16/20110319135224]]
[20:31:50] -!- newbynobi has quit [Quit: Verlassend]
[20:35:42] <slime2k> http://amalas.eu/~slime/PROJECTS/CNC/heimatstube/P1010086.JPG
[20:35:45] <slime2k> http://amalas.eu/~slime/PROJECTS/CNC/heimatstube/P1010088.JPG
[20:35:48] <slime2k> http://amalas.eu/~slime/PROJECTS/CNC/heimatstube/P1010090.JPG
[20:37:29] <alex_joni> looks great
[20:37:38] <cradek> neat. how long does that take to cut?
[20:41:04] <slime2k> hmmm
[20:41:14] <slime2k> hard to say
[20:43:42] <slime2k> did the roughing yesterday evening
[20:43:55] <slime2k> about 3-4 hours
[20:44:13] <slime2k> and finishing today
[20:44:41] <slime2k> because of the bug every letter in a single file :)
[20:45:03] <fragalot> shiny
[21:13:10] -!- distracted has quit [Ping timeout: 260 seconds]
[21:14:52] -!- acemi has quit [Quit: WeeChat 0.3.2]
[21:16:16] -!- micges has quit [Quit: Ex-Chat]
[21:16:16] -!- distracted [distracted!~distracte@] has joined #emc
[21:23:14] -!- Valen [Valen!~Valen@ppp121-44-184-217.lns20.syd7.internode.on.net] has joined #emc
[21:24:20] -!- andypugh has quit [Quit: andypugh]
[21:43:32] -!- andypugh [andypugh!~andy2@cpc2-basl1-0-0-cust1037.basl.cable.virginmedia.com] has joined #emc
[22:10:04] -!- kb8wmc has quit [Quit: ChatZilla [Firefox 3.6.16/20110323162424]]
[22:16:10] -!- cncbasher has quit [Quit: KVIrc Insomnia 4.0.0, revision: 3900, sources date: 20100125, built on: 2011-03-15 15:32:03 UTC http://www.kvirc.net/]
[22:20:50] -!- micges has quit [Quit: Ex-Chat]
[22:44:33] <elmo40> interesting... this guy submitted his cnc machine to the Dyson awards. http://redbluecnc.blogspot.com/
[22:44:33] -!- jdhNC [jdhNC!~jdh@miranda.org] has joined #emc
[22:44:39] <elmo40> kinda lame.
[22:44:46] <elmo40> but universal
[22:45:03] <elmo40> nothing explained about the computer that drives it!
[22:50:25] <chille> LOLHAX!!!!
[22:50:29] <chille> IT'S MINE!!!!
[22:50:32] <chille> it's fucking mine :D
[22:50:38] <chille> i just bought a ASEA IRB 6 robot
[23:08:14] -!- houseparty [houseparty!~mint@adsl-99-37-241-24.dsl.sbndin.sbcglobal.net] has joined #emc
[23:10:07] -!- robh__ has quit [Ping timeout: 264 seconds]
[23:17:29] <andypugh> Sounds like fun
[23:29:15] <danimal_garage> hi
[23:32:35] -!- logger[psha] has quit [Ping timeout: 260 seconds]
[23:33:55] -!- logger[psha]_ [logger[psha]_!~loggerpsh@] has joined #emc
[23:35:29] -!- nomicon_ [nomicon_!~mo@qnix.se] has joined #emc
[23:35:32] -!- Vq_ [Vq_!~vq@90-227-210-39-no122.tbcn.telia.com] has joined #emc
[23:35:37] -!- theorbtwo has quit [*.net *.split]
[23:35:37] -!- OoBIGeye has quit [*.net *.split]
[23:35:37] -!- cjdavis has quit [*.net *.split]
[23:35:37] -!- Vq has quit [*.net *.split]
[23:35:37] -!- logger[psha] has quit [*.net *.split]
[23:35:38] -!- nomicon has quit [*.net *.split]
[23:35:43] theorb is now known as theorbtwo
[23:36:26] -!- OoBIGeye [OoBIGeye!~kent@81-231-96-193-no48.tbcn.telia.com] has joined #emc
[23:38:58] -!- distracted has quit [Quit: Leaving]
[23:41:51] <elmo40> moo?
[23:44:57] <andypugh> Night all
[23:45:01] -!- andypugh has quit [Quit: andypugh]
[23:47:23] -!- EDocTooR_ [EDocTooR_!~EDocTooR@206-248-162-24.dsl.teksavvy.com] has joined #emc
[23:51:41] -!- EDocTooR has quit [Ping timeout: 252 seconds]
[23:51:43] -!- anonimasu has quit [Ping timeout: 252 seconds]
[23:51:44] -!- anonimasu [anonimasu!~an0n@] has joined #emc
[23:51:44] -!- jared has quit [Ping timeout: 250 seconds]
[23:51:45] -!- CIA-8 has quit [Ping timeout: 252 seconds]
[23:51:45] -!- steves_logging has quit [Ping timeout: 250 seconds]
[23:51:58] -!- sarariman_seb has quit [Ping timeout: 250 seconds]
[23:52:11] -!- kasuga5 has quit [Ping timeout: 252 seconds]
[23:52:28] -!- sarariman_seb [sarariman_seb!~seb@] has joined #emc
[23:52:33] -!- houseparty has quit [Ping timeout: 252 seconds]
[23:52:33] -!- Connor has quit [Ping timeout: 252 seconds]
[23:52:33] -!- lilalinux has quit [Ping timeout: 252 seconds]
[23:52:47] -!- jared [jared!~jordana@pool-108-17-82-7.pitbpa.fios.verizon.net] has joined #emc
[23:52:48] -!- houseparty [houseparty!~mint@adsl-99-37-241-24.dsl.sbndin.sbcglobal.net] has joined #emc
[23:52:56] EDocTooR_ is now known as EDocTooR
[23:53:11] -!- kasuga5 [kasuga5!~kasuga5@wlnt-02-042.dsl.netins.net] has joined #emc
[23:53:54] -!- lilalinux [lilalinux!znc@] has joined #emc
[23:55:03] -!- Connor [Connor!~Connor@] has joined #emc