#emc | Logs for 2009-08-21

[00:03:06] <tom3p> fenn: thx, i was trying to calc the individual command voltages to amps ( and hoping to see the voltage for the A axis that moved 0.8 surface inches past the tooltip in the same time) I dont understand what you said but will chew on it ;) thx
[00:10:22] <fenn> i thought PID did that for you
[00:10:29] <fenn> oh well
[01:22:08] <eric_unterhausen> thinking about buying one of the Intel Atom boards for EMC, any new experience here that would suggest doing otherwise?
[01:25:21] <SWPadnos> I think I had some overheating problems, but I was operating the systems in the sun, in black cases
[01:25:55] <SWPadnos> the machines wouldstop doing anything, then pick up where they left off a minute or two later, which sounds a lot like a thermal shutdown thing Intel has in all their recent CPUs
[01:27:07] <eric_unterhausen> yeah, seems hard to get away from that
[01:27:39] <eric_unterhausen> although my mill is in the basement, so heat isn't usually a problem
[01:27:40] <SWPadnos> adequate cooling or not operating in the sun would probably take care of a lot of it :)
[01:27:44] <SWPadnos> yeah
[01:32:09] <eric_unterhausen> review mentions problems with the fan dying
[01:33:03] <SWPadnos> a lot of people say to replace it immediately, though I haven't
[01:33:19] <eric_unterhausen> what can you use?
[01:33:23] <SWPadnos> also, it might not hurt to add a CPU fan - the one on the board is on the chipset (which takes 2x the power of the CPU)
[01:33:48] <SWPadnos> err, M350, from mini-box.com
[01:33:56] <SWPadnos> lets you plug in a PCI card even
[01:34:07] <SWPadnos> oh, actually I have the M350 LCD
[01:34:26] <SWPadnos> oops, M300 LCD
[01:34:29] <jymm> What would be a preofessional term for "preventing from falling through the cracks", "dont drop the ball", etc?
[01:34:35] <SWPadnos> the M350 doesn't have the PCI slot, I think
[01:34:44] <SWPadnos> "attention to detail"
[01:35:03] <SWPadnos> jymm, ^^
[01:35:49] <jymm> SWPadnos: I was considering more customer service oriented. Like when you're told "welll call you back" but never happens.
[01:36:03] <SWPadnos> "excellent follow-through"
[01:36:26] <jymm> But is "follow through" the right veribage?
[01:36:32] <SWPadnos> I think so
[01:37:26] <SWPadnos> "meets or exceeds customer expectations, with attention to detail and 100% follow-through"
[01:37:31] <jymm> http://en.wiktionary.org/wiki/follow_through
[01:37:41] <SWPadnos> is that enough BS^H^H management-speak for you?
[01:38:16] <eric_unterhausen> b sub sigma
[01:38:21] <SWPadnos> exactly - to finish; to complete, especially, of ac ommitment
[01:38:42] <SWPadnos> oh. you're right - I didn't mention five "9"s
[01:39:13] <jymm> It's what I would use in casual conversation, or informal, but just doens't sit right with me for some reason. very very close though
[01:39:40] <eric_unterhausen> use Hindi then
[01:39:42] <tom2> fenn: i see how pid can handle an error but some other code must divvy up the error into dx dy dz and hand those to individual pid's. isnt there one pid for each joint?
[01:40:24] <SWPadnos> tom2, yes, PID is per joint, not on the "pose" as a whole
[01:43:57] <jymm> SWPadnos: "Strong dedication to customer communications and follow through."
[01:44:06] <jymm> ?
[01:44:11] <SWPadnos> commitment, not dedication
[01:44:26] <SWPadnos> either "dedicated to ...", or "strong commitment to ..."
[01:44:34] <eric_unterhausen> a pig is committed, chicken is dedicated
[01:44:48] <jymm> * jymm lol @ eric_unterhausen
[01:44:56] <SWPadnos> indeed
[01:45:03] <jymm> eric_unterhausen: What about a TurDuckIn ?
[01:45:52] <jymm> eric_unterhausen: http://en.wikipedia.org/wiki/Turducken
[01:46:37] <eric_unterhausen> look up a recipe for cooking a camel
[01:46:52] <jymm> no thanks, heh
[01:47:06] <eric_unterhausen> first you cook 30 turkeys
[01:48:17] <jymm> turducken in a pig, in a deer, in a cow, in a shark, in a hippo, etc
[01:48:45] <jymm> wait, toss in there goat before pig I think
[01:49:33] <eric_unterhausen> Is the 945GCLF2D better than the 945GCLF2 ?
[01:50:14] <SWPadnos> one of them has the S-video port, the other doesn't
[01:50:24] <SWPadnos> I don't remember which, but I think the more expensive one has it
[01:50:34] <jymm> I think the 2D
[01:50:36] <eric_unterhausen> they cost the same at newegg
[01:50:46] <SWPadnos> oh, then get the (one with the port)
[01:51:13] <eric_unterhausen> ok, d doesn't have port
[01:51:26] <eric_unterhausen> d stands for "doesn't"
[01:51:33] <SWPadnos> maybe the D doesn't have it
[01:51:35] <SWPadnos> yep
[01:51:44] <SWPadnos> I'm pretty sure mine are non-D, and they have it
[01:52:03] <SWPadnos> note that neither has a DVI connector though, so LCDs will look shitty
[01:52:08] <eric_unterhausen> It's important to be able to watch tv on my mill
[01:52:23] <SWPadnos> of course
[01:52:29] <SWPadnos> on a TV, that is
[01:52:35] <jymm> get a hdhomerun then =)
[01:52:38] <SWPadnos> (you still can on a VGA montir
[01:52:40] <eric_unterhausen> I find the a/d converters to give acceptable quality
[01:52:40] <SWPadnos> monitor
[01:52:47] <eric_unterhausen> on lcds
[01:52:52] <SWPadnos> it's the synch that gets me
[01:53:18] <SWPadnos> I often see pixel bleed or other artifacts on LCDs driven from analog sources
[01:54:58] <eric_unterhausen> I guess that's the advantage of having screwed up eyes
[01:58:17] <eric_unterhausen> ddr2 800 should work, correct?
[01:59:44] <SWPadnos> 667 will work. I'm not sure the board goes to 800
[02:00:34] <SWPadnos> actually, it's 533
[02:00:49] <SWPadnos> the tech manual says that DDR2 667 and 800 will work, but at 533 speed
[02:01:13] <eric_unterhausen> newegg says 533 fsb, uses 667 memory
[02:01:50] <SWPadnos> the tech manual says "DDR2 533 MHz or DDR2 400 MHz SDRAM DIMMs (DDR2 800 MHz and DDR2 667 MHz validated to run at 533 MHz only)"
[02:02:34] <SWPadnos> I haven't looked to see if mine are running at 667 or not
[02:02:46] <SWPadnos> there could have been an update or something
[02:03:19] <eric_unterhausen> no, I think it runs at 533/400, they are being weird about memory
[02:05:08] <eric_unterhausen> d now stands for "disappeared"
[02:12:37] <cradek> yay, jr is moving again
[02:13:14] <cradek> also, I much prefer tuning pid loops with halcmd to jumpers and trimpots
[02:19:20] <skunkworks928> heh
[02:20:16] <skunkworks928> eric_unterhausen: with the stick emc kernel - a bit under 20k latency. others have said better using smp.
[02:20:39] <skunkworks928> aprox 18-19us
[02:21:24] <skunkworks928> night time :)
[02:21:27] <SWPadnos> funny - the GOAL3 was much better
[02:21:30] <skunkworks928> or bet dime even
[02:21:34] <skunkworks928> wow
[02:21:34] <skunkworks928> time
[02:21:35] <SWPadnos> but much slower
[02:21:46] <SWPadnos> stick bet dime
[02:21:47] <skunkworks928> yah - the goal board is very good
[02:21:48] <SWPadnos> good night :)
[02:22:14] <skunkworks928> night :)
[07:12:42] <numen> good morning
[12:46:55] <skunkworks_> jepler: great write-up on the zenbot.
[12:49:51] <skunkworks_> jepler - on the right 2 pads - the isolation trace between the pad and the lower copper seems wider on the left pad (of the 2 right most pads) Should they have been the same? wondering if you do have a little bit of backlash in both axis.
[13:20:44] <motioncontrol> good evening.i have a question about the lengh compensation tool when the spindle is orizzontal and move along y axis.is possible enable the lengh compensation along y axis when i select g18 plane? how is possible modification : motion.c ? thanks
[13:22:05] <micges> motioncontrol: there is no tool length comensation along y in emc2
[13:23:16] <motioncontrol> i thing is possible modification, you can help me where is write the code for compensation along z ?
[13:23:50] <micges> comensation is in many files of emc2
[13:24:48] <motioncontrol> i have see before some line type trans --> z ecc..!!
[13:24:56] <jepler> skunkworks_: I see what you mean in that photo
[13:25:36] <motioncontrol> hello jepler
[13:25:54] <SWPadnos> motioncontrol, would you also expect diameter compensation to switch to the XZ plane?
[13:26:04] <jepler> if the system has a single spindle which is horizontal, just designate that axis Z
[13:26:15] <SWPadnos> jepler, it's a two-spindle system
[13:26:19] <SWPadnos> iirc
[13:26:22] <jepler> ok, I didn't read everything
[13:26:31] <SWPadnos> from way back, several days ago
[13:26:32] <motioncontrol> the sistem have two spindle, one vertical and another orizzontal
[13:31:15] <numen> how could i solve the problem with the manual tool change? and the z hight
[13:31:16] <numen> ?
[13:31:49] <jepler> the option to have a W tool offset instead of X is implemented in revisions be142b3 and 563c609
[13:33:26] <SWPadnos> numen, did you look at tool-length-probe.ngc?
[13:33:34] <numen> where to find?
[13:33:48] <SWPadnos> in the nc_files directory of your emc2 install
[13:34:18] <SWPadnos> you can make the measurement part into a subroutine that you call after every tool change
[13:35:48] <jepler> http://git.linuxcnc.org/gitweb?p=emc2.git;a=commitdiff;h=be142b3 http://git.linuxcnc.org/gitweb?p=emc2.git;a=commitdiff;h=563c609
[13:35:59] <numen> is it possible, too, that this works automatic?
[13:36:02] <jepler> adding a Y offset would be substantially similar
[13:36:43] <SWPadnos> numen, I don't know of a way of making it work totally automatically, with only TxxM6 code in the program
[13:37:08] <SWPadnos> if you can make your post emit TxxM6 then O<probe> call, then it will be effectively automatic
[13:37:22] <SWPadnos> (assuming you make the probe code into a subroutine called "probe")
[13:38:36] <numen> i dont understand
[13:39:10] <numen> language problem ^^
[13:39:21] <SWPadnos> I don't think you can make "Txx M6" do the probing automatically
[13:39:30] <numen> this i understood
[13:39:45] <SWPadnos> it is possible to make it work without the operator having to do anything after inserting the new tool
[13:40:00] <SWPadnos> but you have to add some lines to your G-code to do the probe
[13:40:03] <numen> that sounds good
[13:40:14] <numen> and how will i have to do this?
[13:40:37] <SWPadnos> read and understand how the probing and offsets work in tool-length-probe.ngc
[13:40:50] <SWPadnos> then put the actual probing code into a subroutine
[13:41:16] <SWPadnos> name the file the same as the subroutine, and put that file in your nc_files directory
[13:41:34] <SWPadnos> then add a line after every tool change that calls the subroutine
[13:42:06] <SWPadnos> so if you call the subroutine "probe", like this:
[13:42:14] <SWPadnos> O<probe> sub
[13:42:19] <SWPadnos> ... (the probing code)
[13:42:26] <SWPadnos> O<probe> endsub
[13:42:35] <SWPadnos> then you should name the file probe.ngc
[13:42:46] <SWPadnos> and you should call it by adding this line after every tool change:
[13:42:51] <SWPadnos> O<probe> call
[13:43:25] <numen> for what is the O? does it need?
[13:43:42] <SWPadnos> O is the g=code word we use for subroutines
[13:43:48] <SWPadnos> g-code
[13:43:57] <SWPadnos> you should read about that too :)
[13:44:20] <numen> so all what is in thät tool-length-probe.ngc i will have to add to probe.ngc?
[13:44:53] <numen> and where can i define, where the switch is?
[13:44:56] <SWPadnos> I think it's the parts where it says (measure reference tool) and (measure next tool)
[13:45:18] <SWPadnos> I'm not sure about the specifics
[13:45:21] <numen> i will have to make 2 subrutines?
[13:45:36] <SWPadnos> that is possible, but one should only get run once per program
[13:45:47] <SWPadnos> the other after every tool change
[13:46:06] <numen> i cant add this in 1file?
[13:46:36] <SWPadnos> you can add the subroutines to every G-code file if you want to
[14:03:37] <numen> but if i create i file called Oreftool, there i add the code unde the measure reference tool
[14:04:38] <numen> then before the first tool change i add in my gcode Oreftool
[14:04:41] <numen> is that right?
[14:05:45] <numen> then i create a second file called Oprobe.ngc there i add everything which is under measure next tool and in gcode i always add Oprobe after toolchange?
[14:06:18] <numen> is that correct?
[14:06:28] <SWPadnos> that's the right idea
[14:06:34] <SWPadnos> I think that you don't need the O in the filename
[14:06:55] <SWPadnos> you would call the file with O<reftool> in it "reftool.ngc"
[14:07:08] <numen> ok
[14:07:19] <SWPadnos> I think. you should check the manual though, I don't know the specifics
[14:07:28] <numen> and when the tool connects to the switch, it should go to gnd or to vcc?
[14:07:43] <numen> but where can i define, where my probe is?
[14:09:49] <skunkworks_> numen: look at g28
[14:09:50] <skunkworks_> http://linuxcnc.org/docs/html/gcode_main.html#sub:G28,-G30:-Return
[14:10:02] <skunkworks_> or 30
[14:11:40] <skunkworks_> also look the tool-length-probe.ngc sample program (it uses g30 for the switch location)
[14:12:29] <SWPadnos> That's what he's been looking at. It's just not set up as subroutines you can call - it's more of a demonstration of the method
[14:15:11] <skunkworks_> right - but if you set the switch location to g28 - it could be called in the subrutine
[14:16:51] <SWPadnos> gotta run. bbl
[14:24:34] <skunkworks_> fenn: I wonder for something cheap and dirty how automotive tie-rod ends would work.
[14:24:54] <skunkworks_> screw them into each end of your actuator.
[14:27:40] <skunkworks_> I visulize them going into the edge of the top and botom plates.
[14:29:30] <numen> skunkworks_ but is this a maschin coordinate or a workpiece?
[14:30:35] <skunkworks_> machine coordinate
[14:30:58] <skunkworks_> so if your machine homes correctly - it is always in the same place :)
[14:32:23] <numen> could it be negative for y, too?
[14:34:11] <skunkworks_> If your machine can go there - yes.
[14:34:34] <numen> but the end switch?
[14:35:27] <numen> is it possible, that it ignores the end switch for tool probing?
[14:36:53] <numen> but with keyboard i cant go to y-20
[14:36:58] <numen> it stops on y0
[14:37:39] <numen> or do i have to change to homing value?
[14:45:24] <numen> and where will i have to define the G30 coordinates?
[14:46:31] <skunkworks_> well - you can either edit your var (5181-5186) file - or set it using g30.1
[14:46:48] <cradek> you mean using MDI #5181=...
[14:47:13] <cradek> but g30.1 is easiest
[14:47:17] <skunkworks_> or that :)
[14:48:48] <alex_joni> http://imgur.com/gallery/XnQUC
[14:50:12] <skunkworks_> funny
[14:54:17] <alex_joni> http://www.theregister.co.uk/2009/08/21/chrome_64_bit_linux/ <- interesting
[14:58:25] <numen> skunkworks928 but this is just for x and y axis?
[14:58:30] <numen> or z, too?
[14:58:48] <skunkworks_> all axis if you want
[14:58:52] <geo01005> numen, it does all 9 axis.
[15:01:04] <numen> do i have to define z, too?
[15:01:13] <numen> to z0 or something like this
[15:05:30] <alex_joni> http://notdoppler.com/multitask.php <- ROFL
[15:14:06] <seb_kuzminsky> iphone cad to 3d printer :-) <http://www.instructables.com/id/Creating-a-Model-for-3D-Printing-on-an-iPhoneiPod/>
[15:21:51] <holy> "3d printer"?
[15:31:18] <numen> geo01005 should i set all axis?
[15:33:54] <archivist> numan you are doing whatever, you should know
[15:35:43] <geo01005> You have a 3axis mill.
[15:36:29] <geo01005> To set the g30 position, just put the machine in the right position then go into MDI mode and type "G30.1"
[15:36:52] <geo01005> That should store the value of all the axis (your machine only uses 3 of them though)
[15:38:22] <geo01005> Then I image when you want to go to the tool probing position you would want to move up in the Z axis prior to moving the X or Y axis.
[15:38:40] <geo01005> So first "G30 Z0"
[15:38:51] <geo01005> Then "G30 X0 Y0"
[15:39:47] <geo01005> Then use G38 to do the straight probing.
[15:40:31] <geo01005> Somebody correct me if I wrong. And numen, you will have to figure out all the details.
[15:40:45] <cradek> I think you mean g0 g91 g30 z0
[15:41:06] <cradek> g90 z0 might be down, which is probably not what you want
[15:41:34] <cradek> I prefer the form g0 g53 z0 for "go up" because it doesn't make you mess up your g90/g91 setting
[15:45:11] <geo01005> Yes the G53 would work better wouldn't it.
[15:46:32] <jepler> sometimes I think a "non-modal incremental code" would be nice, for reasons like this
[15:46:48] <geo01005> However, if you just did a tool change, you are probably at a safe rapid Z level, in which case you would want to move in X and Y before making any moves "down" toward a tool length probe... I think.
[15:51:07] <geo01005> Hmm, this is making me want to make a probe for my machine :)
[15:52:06] <holy> probe is an electrical device for shorting and stopping at the very same moment, isn't Pt?
[15:52:43] <archivist> probing for some will be complex, think centerline of a gear cutting tool and its od
[15:58:49] <numen> i think, i will just mount a short alu part, then i will give about 5Volts on my tool and touch the plate
[15:58:53] <numen> quite easy
[15:59:39] <archivist> if rotating this will cut
[16:00:08] <holy> Should it be a switch, which turns off the moves?
[16:00:09] <numen> why rotating? not neceserry
[16:00:38] <SWPadnos> holy, in that case, how would you move the tool off the sensor?
[16:00:43] <archivist> if not rotating how are you going to know peak extent
[16:00:44] <SWPadnos> if the motors are turned off
[16:01:54] <holy> SWPadnos: simply, when it gets shorted, you see the coordinetes in the var.tbl file.
[16:01:57] <archivist> is pillar not resilient something breaks when you crash into it
[16:02:02] <archivist> if
[16:02:10] <holy> SWPadnos: How didn't you know it? ;)
[16:02:15] <SWPadnos> holy, the way it works is that the motion controller sees the probe input turn on, and stops motion
[16:02:17] <numen> you dont have to go so fast ^^
[16:02:32] <SWPadnos> if you just turned off motor power, you'd have an awfuully hard time moving off the switch
[16:03:22] <holy> SWPadnos: Can't I just measure the distance after changing the tool?
[16:03:39] <SWPadnos> yes. that's unrelated to the method used to stop motion
[16:05:38] <geo01005> For A tool length probe, I would prefer a spring loaded plunger type probe. Doesn't require electrical contact, and allows the tool to move farther than where the switch trips.
[16:05:42] <holy> SWPadnos: when I see the "Insert tool #XXX and continue" notification, how could I tell the program the length of the tool?
[16:05:53] <geo01005> This allows for faster measurement.
[16:06:18] <SWPadnos> read back, we've been talking about tool length probing all day
[16:07:00] <holy> ok
[16:09:00] <geo01005> I made a plunger type tool setter a few years ago, worked fairly well (repeatable to about .001") until some students started dropping it on the concrete for fun.
[16:09:51] <archivist> 1 thou can be a bit barn door
[16:10:27] <geo01005> yes, but for routine junk it is ok.
[16:10:39] <geo01005> I certainly could have made it better.
[16:12:23] <geo01005> However to get a really good probe, you might as well buy a good one.
[16:31:19] <holy> Last time I tried it, subroutine didn't handle parameters like #1 or #<_z-rough-tolerance>
[16:40:11] <eric_unterhausen> I blew up a relay last night, but it turns out it was the socket that failed. relay itself still works
[16:47:12] <motioncontrol> good evening.in sim configuration i have create another thread :loadrt threads name1=plc period1=10000000 because in real maschine i have great plc classicladder(upper 100rung) and first i refresh classicladder on servo-therad 1 .now i want create another thread in real maschine, but have a error because i have now 3 thread :loadrt [EMCMOT]EMCMOT base_period_nsec=[EMCMOT]BASE_PERIOD servo_period_nsec=[EMCMOT]SERVO_PERIO
[16:47:12] <motioncontrol> D traj_period_nsec=[EMCMOT]TRAJ_PERIOD key=[EMCMOT]SHMEM_KEY num_joints=[TRAJ]AXES
[16:47:12] <motioncontrol> . i want delete traj thread, i don't use this thread.i see another configuration and is write oadrt [EMCMOT]EMCMOT base_period_nsec=[EMCMOT]BASE_PERIOD servo_period_nsec=[EMCMOT]SERVO_PERIOD num_joints=[TRAJ]AXES. in my configuration whith 5i20 card this therad is complete for fuction emc2 ? thanks
[16:50:15] <micges> motioncontrol: what error do you have ?
[16:53:41] <motioncontrol> i have a error on thread not can create because 3 thread there are in my real configuration.for prove i create a new thread in sim condiguration and fuction ok.the question is: my loadt rt in hal file :loadrt [EMCMOT]EMCMOT base_period_nsec=[EMCMOT]BASE_PERIOD servo_period_nsec=[EMCMOT]SERVO_PERIOD traj_period_nsec=[EMCMOT]TRAJ_PERIOD key=[EMCMOT]SHMEM_KEY num_joints=[TRAJ]AXES
[16:53:41] <motioncontrol> is ok or i can modification it for delete for example traj thread and after create a new thread.i read on doc the max number the thread is 3?
[16:55:18] <micges> motioncontrol: pastebin.ca your ini and hal file please
[16:57:55] <motioncontrol> micges, http://pastebin.ca/1537874
[16:59:33] <micges> ok and hal file?
[17:01:18] <alex_joni> darn, can't make more than 55 on multitask
[17:01:46] <motioncontrol> micges, http://pastebin.ca/1537878
[17:04:56] <micges> motioncontrol: I can't see any lines with loadrt threads
[17:05:51] <motioncontrol> micges, yes excuse i send now
[17:07:23] <motioncontrol> micges, http://pastebin.ca/1537885
[17:09:01] <micges> motioncontrol: you must load threads in order from fastest (50000) to lowest (10000000), so you must load plc thread after loadrt motion
[17:10:27] <micges> motioncontrol: there is no limit in threads count in hal, so you can better left traj-thread
[17:10:50] <motioncontrol> micges, http://pastebin.ca/1537889 is ok?
[17:11:15] <micges> should be, try it
[17:13:44] <motioncontrol> ok i thing this is my error , tomorrow i prove on real maschine , more, more...thanks micges.first when the refresh the classic ladder have on servo-thread 1 i have problem with realtime time exceted, becasue my plc classicladder is complex and want 150000ns for complete read .more thanks
[17:13:58] <motioncontrol> micges,
[17:14:54] <micges> sure
[17:14:56] <motioncontrol> micges, one question: but traj thread what is ?
[17:15:25] <motioncontrol> micges, is necessary for calculate the point the trajctory?
[17:15:56] <micges> yes
[17:17:07] <motioncontrol> micges, more thanks can i have another question for you for tool lengh compensation along y axis? thanks for your patience
[17:17:48] <micges> ask
[17:20:26] <motioncontrol> i want modification in source code the tool lengh.now fuction only on z axis, but i have one maschine with 2 spindle.one vertical and another orinzontal.my z axis is vertical,and the tool lenght comp. funtion ok along y. when fuction the spindle orizzontal i change the plòane at g18, but the tool lenght compensation not applicated on y axis, but only z .i want modification the source for example command.c, but i don't
[17:20:26] <motioncontrol> are secure ?
[17:21:25] <motioncontrol> micges, error write funtion ok along y --> fuction ok along z
[17:21:32] <motioncontrol> yes
[17:22:34] <motioncontrol> micges, i have write on emc-development before and the people write me the emc2 non fuction comp. lenght along y
[17:23:23] <micges> motioncontrol: yes there is no tool length comp on axis y
[17:23:45] <micges> it is hard to add this
[17:25:06] <motioncontrol> micges, i thing when the plane is g17 the tool comp. lengh is along z , in g18 the tool comp lenght is along y.you have the source code file for modification?
[17:26:43] <motioncontrol> micges, ok i have traslate now .excuse for my english.i thing it is important for complete cnc
[17:28:39] <cradek> the tool length compensation does not and will not move with the g17/g18 plane. doing that makes no sense whatsoever.
[17:28:58] <micges> it isn't in emc2 because noone required it so far
[17:29:38] <cradek> tool length offsets and arc planes have nothing to do with one another
[17:30:15] <micges> motioncontrol: you can add feature request on www.sf.net/projects/emc
[17:30:20] <micges> bbl
[17:31:01] <archivist> his second spindle is at 90 degrees
[17:31:31] <jepler> you can add a feature request, but that does little to actually cause a feature to be implemented.
[17:31:54] <cradek> I understand. but that has nothing to do with arc planes either.
[17:31:59] <motioncontrol> ok i undertsand , but for maschine with 2 spindle the emc not is can used
[17:32:37] <motioncontrol> the arc plane change ok with g17-->g18 is only tool lenght not change z-->y
[17:33:27] <motioncontrol> ok i add the request for this improvment i thin is important for emc to be complete
[17:34:08] <jepler> *shrug* new features need people to implement them. submitting a feature request doesn't achieve that.
[17:35:06] <motioncontrol> thanks for all support.i want contribute at improvment the emc with donation money.exist a bank support for payment my contribute for people development?
[17:35:07] <archivist> often the person needing it most implements it
[17:36:42] <cradek> I guess I would like it if tool offsets also had Y. But that is kind of hard with the current tool table.
[17:36:53] <cradek> I would happily review a patch that solves this problem
[17:37:26] <archivist> I think a few more fields to the tool table too
[17:38:13] <motioncontrol> cradek, i thing create a new variable for y table not use the z table
[17:38:14] <cradek> some day we will need to move away from the human-editable text file style of tool table because of these problems
[17:38:36] <archivist> I agree
[17:38:37] <cradek> motioncontrol: yes the Y tool length offset would be separate. each tool would have x,y,z offsets.
[17:38:46] <geo01005> Could that problem be solved with a kinematics module?
[17:39:08] <cradek> geo01005: no, because kinematics doesn't know what part of the current pose is caused by tool offsets
[17:39:47] <geo01005> I'm thinking of just pretending that the second spindle is also in the Z axis direction.
[17:39:59] <geo01005> Writing the g-code to account for that.
[17:40:09] <motioncontrol> cradek, you have the modification now or i wait for improvment on new version the emc2.i want consigne my maschine complete.thenks
[17:40:11] <geo01005> Then have a hal pin that switches between the two spindles.
[17:41:06] <cradek> motioncontrol: I am only saying what I think would be necessary for someone to add the feature you want. I am not promising that I will do it now or any other time.
[17:41:45] <motioncontrol> cradek, ok more thanks for your patience
[17:41:51] <cradek> welcome
[17:43:24] <cradek> if your two spindles are in the same plane, you could say one is along Z and the other is along X, and you would have the necessary tool offsets
[17:43:59] <cradek> if they are not in a plane, you need all three offsets, and you would need to change emc.
[17:45:37] <geo01005> So you don't think that it would be possible to use a kinematics transformation to switch what is normally the X axis to being a virtual Z axis?
[17:45:55] <alex_joni> you can do that, but that doesn't really help
[17:46:06] <cradek> kinematics don't have access to tool length offsets, so you can't move an offset from one place to another
[17:46:12] <alex_joni> offsets are done in the interp/canon, way long before they reach motion, where you do kinematics
[17:46:14] <numen> i think, i work too much
[17:46:56] <geo01005> I assumed that the tool offsets are done prior to the kinematics module.
[17:47:09] <alex_joni> yes, way before
[17:47:34] <numen> 15hours a day are too much...
[17:48:47] <geo01005> I was thinking of two separate configurations on the same machine.
[17:48:59] <jepler> geo01005: oh, you mean have kinematics rotate the coordinate system when you change spindle so the current spindle is always "Z"?
[17:49:08] <geo01005> Yes :)
[17:50:10] <geo01005> Then you wouldn't have to do anything different with any of the tool compensations.
[17:50:26] <cradek> I don't think that is true
[17:51:22] <cradek> if the spindles are not planar, you need three offsets. There are not three offsets. Renaming axes gives you no benefit.
[17:51:37] <cradek> If the spindles ARE planar, call them X and Z, and you don't have to rename axes.
[17:52:04] <jepler> and there's the issue of the discontinuty in coordinates when you change spindles at locations other than machine 0,0,0
[17:52:20] <geo01005> Hmm, I
[17:52:36] <geo01005> I'm not seeing these problems, but I'm often blind to problems with my Ideas.
[17:54:14] <geo01005> I see the discontinuity, but that could be accounted for right?
[17:54:53] <cradek> no, because the position interpolator is before kinematics. if kinematics causes a discontinuity, it is a position jump -- it is too late to interpolate along it
[17:55:41] <cradek> that is why things like offsets are in the interpreter. If you change one, the position of the machine doesn't move. It just gets applied next time you make a move.
[17:55:43] <geo01005> Perhaps this would be better done in the hal with some RT components.
[17:55:55] <cradek> no, same problem
[17:59:33] <geo01005> hmm, I see what you are saying now.
[18:00:08] <geo01005> had to write down some numbers to see why.
[18:04:53] <motioncontrol> ok i have save on souceforge the request on new tracker for tool compensation lenght .thanks at all.
[18:05:56] <geo01005> what about a virtual axes (I'm not sure if it would be A,B,or C) That would simulate the tool moving from being along the Z axis to being along the X axis?
[18:06:05] <motioncontrol> geo01005, excuse for my question.Peter at mesa speack me about the spi bus. i have hardware , but not have the idea for complete system. you have news?
[18:06:37] <geo01005> No, there hasn't been any work done on SPI yet.
[18:07:57] <geo01005> seb_kuzminsky, am I right about that?
[18:08:37] <motioncontrol> geo01005, ok thanks
[18:09:02] <seb_kuzminsky> yeah no work on it yet
[18:09:05] <seb_kuzminsky> soem day
[18:10:21] <geo01005> cradek, or Jepler, Is that Idea of a virtual C axis any good?
[18:10:44] <cradek> do you mean work rotation in the XY plane?
[18:10:55] <cradek> (which we already have)
[18:11:42] <geo01005> Hmm maybe rotation about the Y axis.
[18:11:54] <geo01005> But yes.
[18:12:10] <cradek> we already have coordinate system rotation around Z only
[18:12:52] <geo01005> No, no, I was thinking of pretending that the machine had another axes that rotates about the Y axis.
[18:13:06] <cradek> maybe you meant B then, not C
[18:13:31] <geo01005> Sure, I need a diagram or good memory to tell which one.
[18:14:49] <geo01005> Then you could position the virtual axis so that the tool orientation was correct right ?
[18:30:39] <BJT-Work> * BJT-Work takes the long way home today
[18:34:07] <alex_joni> geo01005: it only works with special kins I think
[18:34:18] <alex_joni> and applying the offsets to UVW
[18:34:45] <geo01005> I was just trying it out with 9axis sim.
[18:35:08] <geo01005> does he need cutter compensation?
[18:37:20] <alex_joni> yeah
[18:37:22] <geo01005> hmm, wonder why cutter compensation dosen't work while the YZ plane is selected....
[18:37:40] <geo01005> I know that doesn't have sense for a 3 axis mill.
[18:37:48] <geo01005> But I though it was supported in general.
[18:37:51] <cradek> that's why :-)
[18:38:12] <cradek> emc2 does mill cutter comp in g17 and lathe cutter comp in g18
[18:38:29] <geo01005> I see, just never been needed in the other plane.
[18:38:43] <cradek> yeah, it hardly makes sense
[18:39:08] <cradek> it might make a tiny bit of sense in some cases if using a ball-end mill
[18:40:15] <geo01005> Well, it make sense for some 5 axis stuff.
[18:40:36] <cradek> yeah 5 axis cutter comp is a whole other story
[18:42:09] <geo01005> well when working on one of the coordinate planes it isn't any harder than cutter comp in the XY plane.
[18:43:14] <geo01005> But why use an expensive 5 axis machine in those cases? It is probably easier to just re-fixture the work.
[18:44:01] <geo01005> Well, ok so cutter comp wouldn't work, but tool length compensation would work.
[18:45:24] <geo01005> I suppose that isn't very helpfull in production settings.
[18:51:30] <skunkworks> http://cnczone.com/forums/showpost.php?p=645342&postcount=27
[18:52:04] <eric_unterhause1> OT: McAffe went from a net worth of $100m to $4m and now is selling properties in order to eat. Nice investment strategy
[18:56:26] <jepler> "I found a bug in CPLDs" is right up there with "I found a bug in gcc" in terms of how often you'd be wrong
[18:56:52] <cradek> "I found a security hole in rm" haha
[18:58:25] <eric_unterhause1> I found a bug in Visual Studio
[18:58:38] <eric_unterhause1> C#
[18:59:40] <skunkworks> looks like it is still un-resolved.
[19:00:47] <skunkworks> guess it has been less than a month
[19:03:28] <jepler> I can't help recalling how hard Jon Elson and I looked for wiring/signal integrity problems yet he eventually discovered the flaw in his fpga code..
[19:03:59] <jepler> at fest on John A's system, that is
[19:04:03] <cradek> was that an index problem?
[19:04:12] <cradek> I don't remember this
[19:04:41] <jepler> steppers with ppmc generating the step pulses
[19:04:47] <jepler> it would ferror at random
[19:04:49] <cradek> oh right
[19:04:53] <cradek> ugh
[19:05:13] <cradek> I remember staying away from that project
[19:05:23] <jepler> I won't fault that choice
[19:08:44] <cradek> anyone know any tricks for unplugging one-shot oil lines?
[19:09:41] <anonimasu> got stuff stuck in them?
[19:10:06] <cradek> some work, some don't - I can only assume they're plugged
[19:10:19] <anonimasu> canister with a hole in the bottom
[19:10:23] <eric_unterhause1> I replaced many of mine
[19:10:32] <anonimasu> and a compressed air input on top
[19:10:40] <anonimasu> and fill it with solvent
[19:11:02] <cradek> eric_unterhause1: how? with what?
[19:11:29] <anonimasu> for the really clogged ones I took some wire and put it through them..
[19:11:34] <eric_unterhause1> are they built in, or actually separate tubes?
[19:11:40] <eric_unterhause1> on a BP, they are tubes
[19:12:08] <cradek> they are tubes
[19:12:10] <eric_unterhause1> expensive though
[19:12:24] <eric_unterhause1> pretty sure msc stocks the stuff you need
[19:12:42] <cradek> I'll look there, thanks
[19:12:55] <cradek> I haven't even taken it all apart - that's the first step.
[19:12:56] <eric_unterhause1> i would think your problem is mostly in the metering units
[19:13:23] <cradek> yeah - wonder if that's the nozzles, or inline
[19:13:26] <skunkworks> jepler: I thought it really did end up being a cable..
[19:13:43] <skunkworks> but I did leave a few days early
[19:15:26] <jepler> skunkworks: the cables did affect how the problem manifested, but after we'd all gone home, Jon Elson write me and said he'd found and corrected a flaw in the fpga too
[19:15:58] <skunkworks> heh - cool
[19:22:22] <alex_joni> for some values of cool
[19:24:20] <skunkworks> ;)
[19:39:36] <eric_unterhause1> cradek: the metering units have a number on them, use the same value or the system will not work right
[20:04:48] <frallzor> toodles
[20:26:14] <Vq> frallzor: god afton
[20:33:08] <frallzor> tut
[20:35:05] <frallzor> http://pici.se/p/nQHGCoqSK/ vad tycks? =)
[20:46:18] <alex_joni> en annan svensk?
[20:46:32] <numen> re
[20:47:16] <frallzor> yes
[20:49:31] <alex_joni> välkommen
[20:50:28] <seb_kuzminsky> alex_joni: jag viste inte att du pratar svenska :-)
[20:50:58] <frallzor> weey massa svenskar här ju :P
[20:51:18] <alex_joni> bara en bit
[20:51:42] <Vq> alex_joni: picked up a few phrases in Skåne have we? :)
[20:52:10] <alex_joni> too few
[20:52:10] <frallzor> * frallzor bor i Skåne
[20:52:27] <alex_joni> frallzor: really? where?
[20:52:35] <frallzor> Hörby
[20:52:41] <celeron55> jag kan prata liten svenska också!
[20:52:44] <Vq> frallzor: not really sweden then :P
[20:52:45] <alex_joni> I was in Lund a couple times
[20:53:04] <frallzor> nice town
[20:53:09] <alex_joni> very nice
[20:53:09] <seb_kuzminsky> Vad är det för skilnad på en Dal mas?
[20:53:33] <frallzor> Dalmas kallar man bara personer man ogillar
[20:53:42] <DaViruz> som mig
[20:54:06] <seb_kuzminsky> Han är brunare på ryggen än på sommaren ;-)
[20:55:13] <alex_joni> logger_emc: bookmark
[20:55:13] <alex_joni> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2009-08-21.txt
[20:56:11] <micges> wow, I drink one beer and this channel is unreadable now :)
[20:56:24] <seb_kuzminsky> haha
[20:56:28] <alex_joni> you need to drink a few so it makes sense again
[20:56:37] <alex_joni> seb_kuzminsky: how come you're speaking swedish?
[20:56:54] <seb_kuzminsky> it's my first language, i was born in sweden :-)
[20:57:03] <micges> alex_joni: ok ;)
[20:57:04] <alex_joni> cool
[20:57:14] <alex_joni> lots of swedes in here
[20:57:19] <alex_joni> (as you probably saw)
[20:57:21] <skunkworks> seb_kuzminsky: really? I don't remember noticing an accent at all.
[20:57:36] <alex_joni> swedish people don't have accents when speaking english
[20:57:43] <alex_joni> or very seldom
[20:57:48] <skunkworks> interesting
[20:57:51] <seb_kuzminsky> alex_joni: no, it's because i'm so suave ;-)
[20:57:57] <alex_joni> (at least that's how I got the impression)
[20:58:01] <jepler> skunkworks: I think if you recall carefully, you'll realize seb doesn't even *look* american
[20:58:03] <alex_joni> seb_kuzminsky: yah, there's that too
[20:58:05] <archivist> swedish chef had an accent
[20:58:14] <seb_kuzminsky> actually my dad's american, and he taught us english growing up
[20:58:17] <skunkworks> heh
[20:58:20] <alex_joni> archivist: he grew up in england though
[20:58:20] <Vq> archivist: wasn't he canadian?
[20:58:38] <archivist> its a puppet!
[20:58:56] <archivist> dunno who did the voice over
[20:59:21] <frallzor> alex_joni people bad at english do =)
[20:59:59] <jepler> wikipedia says: The Swedish Chef is a Muppet that appeared on The Muppet Show. He was operated by Jim Henson and Frank Oz simultaneously and is now puppeteered by Bill Barretta.
[21:00:08] <jepler> but it doesn't say who voiced the chef :( a failure of our collective efforts
[21:02:41] <LawrenceG> http://www.youtube.com/watch?v=mbs64GvGgPU
[21:02:57] <LawrenceG> swedish chef making donuts
[21:03:17] <alex_joni> Jim Henson does the voice
[21:03:18] <skunkworks> heh - the original mad max was dubbed as the distributor didn't think americans would like the austrailian accent.
[21:03:28] <alex_joni> at least that's how I interpret http://muppet.wikia.com/wiki/The_Swedish_Chef
[21:03:47] <seb_kuzminsky> mmm, donut
[21:09:11] <Dave911> Hi guys.... what if anything do you guys use for a Windows IRC client? I've mIRC and it is not too great..
[21:09:35] <archivist> I use a version of xchat
[21:09:48] <alex_joni> I use putty to a linux box, and irssi there
[21:09:56] <Vq> i use putty to a linux box when on windows
[21:09:56] <micges> speaking of languages, have you seen any other Polish people on this channel?
[21:09:59] <archivist> the forked one that does not beg for money
[21:10:16] <alex_joni> micges: what, besides you?
[21:10:25] <Dave911> is xchat better than mIRC ??
[21:10:43] <Dave911> mIRC needs some filters
[21:10:44] <micges> alex_joni: yes
[21:11:05] <Dave911> I'll try that .. thanks
[21:12:09] <Vq> i would probably use erc if i needed a native windows client
[21:12:42] <skunkworks> jepler: I thought seb fit in quite well considering.. ;)
[21:13:00] <Dave911> erc... I'll take a look at that also.. thanks
[21:13:56] <geo01005> I use ChatZilla (Firefox plugin)
[21:16:49] <Dave911> OK, I'll add that to my to try list .. thanks.. :-)
[21:18:50] <skunkworks> I just use the java applet on linuxcnc's site :)
[21:19:01] <skunkworks> but I am lazy that way.
[21:19:41] <archivist> I get better long term connections, logs etc using a native client
[21:26:04] <micges> good night all
[21:40:38] <geo01005> Yeah, he will get it later.
[21:41:13] <geo01005> sorry, wrong place.
[21:43:32] <mozmck> I use pidgin for irc and it works very well. I used to use xchat but I like pidgin better.
[21:44:06] <celeron55> irc client wars! i like irssi the best!
[21:50:17] <AthlonXP1> AthlonXP1 is now known as Dave911
[21:55:09] <Dave911> Yes... client wars ... ;) Chatzilla looks like a winner. I can actually read this one!
[21:56:30] <eric_unterhause1> is there a good linux solution I can use to draw schematics for an EMC system? Or do I have to break out the drafting templates?
[21:57:13] <alex_joni> what kind of schematics?
[21:57:28] <eric_unterhause1> wiring diagrams
[21:57:30] <mozmck> I use kicad, then there's geda, and eagle.
[21:57:40] <alex_joni> right
[21:57:50] <alex_joni> eagle is ok imo
[21:58:10] <archivist> kicad needs users to kick it into shape
[21:58:10] <eric_unterhause1> not sure I want to make library parts for all my boxen
[21:58:32] <alex_joni> archivist: maybe that's why it's called kicad?
[21:58:43] <eric_unterhause1> funny
[21:59:04] <mozmck> It's being worked on all the time, but I agree there are still things that need help.
[21:59:17] <archivist> it seemed lonely so I started the #kicad channel
[21:59:34] <mozmck> I couldn't use eagle because of the board size limitations.
[22:00:02] <skunkworks928> I like eagle.
[22:00:19] <skunkworks928> mozmck: I think we had this discussion at the fest :)
[22:00:28] <mozmck> archivist: do you use kicad?
[22:00:53] <geo01005> I have to wonder sometimes if a cork board with pieces of paper representing components and yarn for connections isn't easiest at first.
[22:01:01] <archivist> mozmck, not often
[22:01:08] <eric_unterhause1> it might be
[22:01:16] <mozmck> skunkworks928: I think it came up :-) I can't afford the expensive programs, and eagle was too limited last time I look.
[22:01:25] <mozmck> looked.
[22:01:44] <skunkworks928> I don't remember the price. so far I have not needed anything bigger. (so far)
[22:01:47] <eric_unterhause1> I really think eda software is not really what I want, probably visio
[22:02:12] <skunkworks928> just remember - traces can go outsid the limit :)
[22:02:22] <mozmck> actually, eagle is one of the more affordable. I think you can get it for around $1000
[22:02:24] <archivist> mozmck, I used to an expensive program from a previous job, so everything I try as a bit odd
[22:03:19] <mozmck> There are a number of people I've seen on the kicad mail list who have switched from expensive stuff.
[22:03:36] <skunkworks928> http://www.cadsoft.de/prices.htm
[22:03:36] <mozmck> I've never used an expensive one so I don't know the difference!
[22:04:18] <eric_unterhause1> I've designed significant circuits with the free version of eagle
[22:04:22] <eric_unterhause1> maybe I should try it
[22:04:45] <mozmck> eric, have you looked at dia?
[22:05:53] <skunkworks928> once you learn eagle - it seems great. it has a bit of a curve
[22:06:13] <mozmck> skunkworks928: looks like you can get a version that would do most anything I would need for around $500
[22:06:37] <mshaver> I've used eagle in the past and now use kicad to make boards. I like kicad!
[22:06:47] <mozmck> I like free better though.
[22:07:56] <mozmck> mshaver: good to hear! One thing they fixed in the last year is zones - they are much better than they used to be.
[22:07:56] <skunkworks928> heh - I should try kicad
[22:08:21] <eric_unterhause1> y
[22:08:22] <DaViruz> eagle is nice - sans the component library
[22:08:31] <DaViruz> it really bothers me how tedious it is to use a package from another lbr
[22:08:40] <eric_unterhause1> sorry, that y was supposed to be for apt-get
[22:08:50] <DaViruz> and that each lbr needs a set of redundant packages
[22:09:56] <mozmck> DaViruz: true. you'd think they would have a standard format.
[22:10:28] <mozmck> components are easy to make in kicad though. but the library handling could be better.
[22:10:40] <DaViruz> i'd like a separate library for packages, and one for symbols
[22:10:53] <DaViruz> which is common for all component libraries
[22:11:06] <DaViruz> but that would make deistribution harder..
[22:11:48] <mozmck> kicad has separate symbol and package libraries, is that what you mean?
[22:12:05] <DaViruz> probably
[22:12:24] <DaViruz> kicad looks interesting
[22:13:07] <mozmck> so I can have one tqfp package an use it for any symbol
[22:13:12] <DaViruz> oh, kicad can use eagle libs
[22:13:32] <DaViruz> or rather convert from eagle libs
[22:13:38] <mozmck> the eagle libs have been converted anyhow.
[22:13:55] <DaViruz> tools for doing it yourself are included
[22:13:59] <mozmck> some of them didn't convert too well I think, but there are a bunch of them
[22:14:04] <DaViruz> oh.
[22:14:29] <mozmck> some are good, and others may have been cleaned up.
[22:14:43] <eric_unterhause1> is the library editor built in?
[22:14:49] <mozmck> yes
[22:15:22] <mozmck> I made symbols for some pic chips, and the longest part was typing in the pin names (80 pins)
[22:16:05] <eric_unterhause1> looks nice, guess I'll have to do a tutorial
[22:16:58] <clip9> wow. kicad really does look nice. I've always used eagle :P
[22:17:18] <eric_unterhause1> I always hated eagle, but at least I could get it to work
[22:17:28] <eric_unterhause1> I was never successful using Cadence
[22:18:30] <eric_unterhause1> of course I didn't find that out until I had a schematic ready to turn into a circuit board
[22:20:18] <clip9> hehe
[22:20:41] <mozmck> I compile my own kicad because the one with ubuntu is kinda old
[22:21:16] <clip9> yeah i just noticed.
[22:21:24] <clip9> 200808
[22:21:29] <clip9> even in karmic
[22:22:03] <archivist> 2007 in hardy
[22:22:10] <mozmck> ah. yeah, I run the cvs version.
[22:22:19] <mshaver> me too - cvs ver
[22:22:26] <eric_unterhause1> I hate compiling from source, are there a lot of dependencies I have to hand compile as well?
[22:23:02] <mozmck> I think wxWidgets is all...
[22:23:08] <clip9> Nah.
[22:23:16] <clip9> just use apt-file
[22:23:44] <clip9> err
[22:23:46] <clip9> auto-apt
[22:23:56] <mozmck> the wxWidgets in Ubuntu may be too old, but I don't remember.
[22:24:18] <skunkworks928> http://electronicsam.com/images/KandT/servostart/schem/latestcurrentlimit/latestboard.png
[22:25:01] <clip9> hm.. eagle?
[22:25:22] <mozmck> looks nice! what value is that cap?
[22:26:14] <skunkworks928> eagle
[22:26:23] <skunkworks928> 1900uf 350v
[22:26:51] <mozmck> eric_unterhause1: you can get a newer version here: http://kicad.sourceforge.net/wiki/index.php/Main_Page
[22:27:04] <skunkworks928> it is a bit messy - but better than the original version :)
[22:30:15] <mozmck> I have some big caps like that - 2400 uF 450v
[22:30:43] <mozmck> maybe I need to make a servo amp!
[22:33:04] <mozmck> Here's where I got them: http://www.bgmicro.com/index.asp?PageAction=VIEWPROD&ProdID=12865
[23:11:00] <skunkworks928> nice!
[23:28:48] <clip9> hmmz. why on earth does kicad scan my whole ~ directory on startup?
[23:36:18] <numen> kicad is strange...
[23:36:22] <alex_joni> /away sleep