#emc | Logs for 2007-11-20

[00:19:48] <eric_U> ot linux question: anyone heard of window bacgrounds turning black possibly when screensaver kicks in?
[00:28:16] <dmess> never noticed... just goes black
[00:50:32] <eric_U> when I come back from the screensaver, the backgrounds of all windows are black, text is screwed up
[00:55:48] <dmess> never seen that...
[06:37:51] <toastydeath> i love it when your only method of contacting an old friend isn't working
[13:45:01] <CIA-23> 03jepler 07TRUNK * 10emc2/src/emc/usr_intf/axis/scripts/axis.py: improve legibility of running line in text area
[13:45:47] <CIA-23> 03jepler 07v2_2_branch * 10emc2/src/emc/usr_intf/axis/scripts/axis.py: backport: improve legibility of running line in text area
[13:46:29] <CIA-23> 03jepler 07v2_2_branch * 10emc2/debian/changelog: note improvement
[13:56:36] <jepler> does anyone here who is not on #emc-devel care about the stuff <CIA> says?
[14:01:31] <skunkworks446> I remember a descusion a while back where the cia was on the #emc channel to show that 'stuff' was getting done :)
[14:01:46] <skunkworks446> maybe it was ray
[14:02:16] <jepler> I remember someone expressing the same sentiment
[14:16:53] <alex_joni> I remember someone, probably Jymmm saying that it's annoying on larger commits
[15:10:17] <skunkworks446> Guest472: Hi
[15:13:03] <Guest472> The new linux kernel has real time support built in. Will Emc2 Bbe using it?
[15:13:35] <SWPadnos> it may be usable for some systems, but probably not for a while
[15:14:30] <SWPadnos> and probably not at all for systems that use the PC for generating step and direction signals
[15:14:33] <Guest472> Is anyone using EMC2 on a pclinuxos setup?
[15:14:38] <jepler> the "real time" support in newer linux kernels is OK for millisecond latency, but not for microsecond latency.
[15:15:14] <jepler> that means that it is promising for systems where a special-purpose board generates the step waveforms, pwm, or whatever; not for systems which generate the step waveform on the PC parallel port
[15:16:15] <jepler> here is the work I did with the "-rt" kernel on ubuntu gutsy gibbon: http://axis.unpy.net/01190912545
[15:16:18] <Guest472> rats
[15:16:29] <jepler> it may apply to other systems where CONFIG_PREEMPT_RT is used
[15:19:22] <Guest472> Is anyone using EMC2 on a pclinuxos setup?
[15:19:37] <jepler> Guest472: I don't know of anyone using it there
[15:21:11] <jepler> mc should work on any system where a real-time kernel is available (this usually means you have to build it yourself) and where the necessary software packages are available, but almost everyone who is using emc uses it on ubuntu linux
[15:21:15] <jepler> +e
[15:22:48] <jepler> the wiki has extensive pages about building the real-time kernel and about building emc. The file debian/control.in in the source distribution lists the ubuntu linux package names that are required to build and run emc, you can use that as a guideline when finding the necessary packages to add to your pclinuxos system
[15:24:05] <Guest472> The ubuntu EMC2 cd wont load on my laptop. Loads for a bit and then hangs on a brown screen!
[15:24:26] <cradek> how much ram?
[15:25:00] <Guest472> 256mb
[15:25:17] <cradek> that's probably not enough.
[15:25:29] <jepler> while the official ubuntu installation notes say 256 megs is enough, many users have had trouble with less than 320MB
[15:25:42] <Guest472> rats
[15:38:16] <skunkworks446> crap - that was the guy on cnczone.
[15:38:25] <skunkworks446> * skunkworks446 got busy
[15:39:35] <rayh> Help with scara under 2.2.1?
[15:40:28] <rayh> All I'm seeing is a display labeled TK
[15:40:56] <alex_joni> rayh: try homing first
[15:43:01] <rayh> There is no axis display only the TK.
[15:43:15] <alex_joni> hmm.. then something's not right
[15:43:16] <rayh> Maybe it's something to do with my install.
[15:43:22] <alex_joni> try running it from the menu
[15:43:27] <alex_joni> err.. frm a terminal
[15:43:32] <alex_joni> (sorry.. a bit distracted)
[15:43:32] <rayh> okay.
[15:43:38] <rayh> brb
[15:45:49] <rayh> emcTrajSetAxis failing
[15:46:07] <alex_joni> check the [TRAJ]AXES =
[15:46:31] <alex_joni> this issue was present in 2.2.0
[15:46:34] <rayh> k
[15:46:37] <alex_joni> has been fixed in 2.2.1
[15:46:43] <alex_joni> mayeb you have an older local config?
[15:47:24] <alex_joni> AXES = 6
[15:47:25] <alex_joni> COORDINATES = X Y Z C
[15:48:14] <rayh> This says Axes = 4
[15:48:26] <alex_joni> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/configs/scara/scara.ini.diff?r1=;r2=1.8;only_with_tag=v2_2_branch;f=h
[15:48:27] <rayh> and C would be 6
[15:48:31] <alex_joni> right
[15:49:59] <rayh> That kills it completely.
[15:51:10] <alex_joni> did you also apply the second fix?
[15:51:17] <alex_joni> AXIS_5 instead of AXIS_3
[15:51:50] <rayh> box is off line so I'm a bit slow
[15:55:51] <rayh> My install says 2.2.1 and I just copied scara config a few minutes ago but it was version 1.8.
[15:56:04] <rayh> Got it running, thanks.
[15:57:06] <alex_joni> maybe you modified /etc/emc2/sample-configs/scara/scara.ini ?
[15:57:24] <alex_joni> that would cause the new 2.2.1 to leave that file alone, and not install a newer version
[15:58:25] <cradek> I think it would ask
[15:59:00] <jepler> it would prompt if you upgrade from the terminal; I'm not sure what it does if you are using the graphical updater
[16:02:01] <skunkworks446> I know that with the gui updater it asked if I want to replace rtapi.conf
[16:02:17] <skunkworks446> (edited for smi)
[16:04:30] <rayh> "All axes must be homed before going into coordinated mode."
[16:04:45] <rayh> I pressed all home.
[16:06:44] <rayh> I suspect that two undefined axes in there are messing up the operation.
[16:07:03] <rayh> I didn't have a 2.2.0 install.
[16:07:16] <rayh> This was new a couple days ago.
[16:07:25] <alex_joni> odd
[16:09:03] <rayh> Yea it is.
[16:13:09] <rayh> Okay. If I add the two intermediate axes and home it does go into coordinated motion mode and let me enter mdi commands.
[16:15:12] <alex_joni> rayh: I can confirm that inside the 2.2.1 package the file is still the old one
[16:15:29] <alex_joni> we'll fix this asap (the next release will probably be this weekend)
[16:19:10] <rayh> np
[16:19:36] <rayh> I get really confusing operation using the axis interface.
[16:19:50] <rayh> After startup in manual mode it will jog each joint.
[16:20:21] <rayh> Switching to MDI seems to put it into world mode but displays are still angular.
[16:20:40] <rayh> and commands seem really odd.
[16:21:35] <alex_joni> you can switch to world with '$'
[16:21:40] <alex_joni> (while still in manual)
[16:28:49] <rayh> okay thanks
[16:41:42] <renesis> http://www.darkertechnologies.com/image/pcb_fix.jpg
[16:41:57] <renesis> i call it "socket head cap screws are sexy"
[16:42:52] <renesis> see where cheap chinese taps ate my fixture
[16:43:11] <skunkworks446> neat.. What is it going to be used for? clamping the edges of a circuit board
[16:43:14] <skunkworks446> ?
[16:43:22] <renesis> yeah
[16:43:28] <renesis> for flipping
[16:43:56] <renesis> theres little nubs underneath the side clamps to locate the board on one Y point
[16:44:06] <renesis> i have 100 boards of the same size
[16:44:31] <skunkworks446> ah. - I have not had that great of luck holding just on the edges.. I use a vaccum table.
[16:44:41] <renesis> how do you drill?
[16:44:58] <skunkworks446> the table is a consumable. (wood)
[16:45:04] <renesis> ooo
[16:45:27] <renesis> i dont like the idea of the pcb drills going thru wood
[16:45:40] <skunkworks446> http://www.electronicsam.com/images/KandT/servostart/top.JPG
[16:46:05] <skunkworks446> oh - it just goes though enough to drill the board.. Maybe .010"
[16:46:47] <renesis> http://www.darkertechnologies.com/image/vap_proto-06.jpg
[16:47:14] <skunkworks446> Cool! :)
[16:47:19] <renesis> thats just 123 blocks, clamps, some parallels
[16:47:36] <renesis> single sided stuff, tho
[16:47:46] <skunkworks446> http://www.electronicsam.com/images/KandT/servostart/mounting.JPG
[16:48:03] <renesis> i mostly want to be able to flip without having to refernce of holes, jiggling the board around
[16:48:11] <renesis> gonna try it out prob thursday
[16:48:25] <renesis> i need to make a piece for the loose bottom clamp assembly
[16:48:32] <renesis> and wed is fucked 10 hour school day
[16:48:41] <skunkworks446> yes - I use pins to register both sides - using g55,g56 coordinates for top/bottom
[16:48:57] <renesis> wow nice
[16:48:59] <skunkworks446> http://www.electronicsam.com/images/KandT/servostart/ampmess.JPG
[16:49:09] <renesis> super to220 w00t
[16:49:15] <skunkworks446> :)
[16:49:45] <renesis> like to-287 or some shit
[16:50:19] <renesis> i want to get some Al extrusions to make chassis sides with
[16:50:48] <skunkworks446> TO-247
[16:51:05] <renesis> oh yeh
[16:51:26] <renesis> confused cuz to-218 i think
[16:51:33] <renesis> .2" pitch thing
[20:35:45] <cradek> rayh: when you use G43 to change tool length, no motion happens. when do you think that change should actually be applied to the tool position?
[20:37:18] <cradek> I'm not asking what emc does, I'm asking what you expect/what you think is right
[20:37:25] <rayh> Hi cradek
[20:37:28] <cradek> hi
[20:37:51] <rayh> Sorry was off trying to kill the scara display after a screw up.
[20:38:32] <rayh> I tend to use something like h43 g0 z0
[20:39:17] <cradek> ok, you put a Z move on the same line as g43, so you avoid the case I'm interested in
[20:39:18] <rayh> sorry
[20:39:23] <cradek> :-)
[20:39:30] <rayh> g43 h1 g0 z0
[20:39:38] <rayh> Right.
[20:39:39] <cradek> I understood
[20:39:59] <rayh> I believe that commercial controllers simply change the displayed position.
[20:40:21] <rayh> Rather than connecting an automatic move to the g43
[20:40:31] <cradek> ok
[20:40:38] <cradek> do you know when they actually make the change?
[20:41:04] <rayh> The next commanded z move I think.
[20:41:23] <rayh> Let me look at a couple of books.
[20:41:37] <cradek> ok so if you move just X, you'd expect Z to remain in the "wrong" (old) location?
[20:41:53] <cradek> that would be great. I can't find anything online (and NGC doesn't explain it)
[20:46:01] <rayh> The mitts 500 example shows g91 g00 g43 h01 z2.0 as a single line
[20:51:48] <rayh> phone brb
[20:58:29] <rayh> I think roltek is off hunting or I'd ask him.
[20:59:06] <cradek> it's ok, it's not life and death or anything
[20:59:21] <cradek> it seems that everyone moves Z right away. I'm interested in what should happen if you don't
[21:00:06] <rayh> I thought that it just changes the position display if you are in relative.
[21:00:22] <cradek> yes it does change the position display
[21:01:15] <cradek> but say you put in a longer tool and then move only X. should Z go up or not?
[21:01:27] <cradek> g49 / g0x0y0z0 / g43h1 / g0x1
[21:04:30] <rayh> I wouldn't think so.
[21:05:21] <cradek> ok
[21:05:38] <cradek> I'm starting to agree. I started off being surprised that it didn't.
[21:06:17] <cradek> (you can make a bunch of moves with Z in the old uncompensated location, which surpises me)
[21:08:07] <rayh> I thought all that it was doing was changing the offset of the current location.
[21:08:31] <rayh> It would be like a g55 on a line by itself.
[21:08:43] <rayh> just shifts the frame of reference.
[21:12:38] <cradek> yeah I see what you mean. If g55 was at x1y1z1, and I did g0x0y0z0 / g55 / g0x1 I wouldn't expect y,z to move
[21:13:02] <rayh> Not until you add a motion command.
[21:13:13] <rayh> I think that is a good way to understand it.
[21:13:36] <rayh> I was looking for my fagor books but I must have loaned them out.
[21:13:45] <anonimasu> cradek: well, if you commands a move for one axis.. others should not move
[21:13:46] <anonimasu> never ever
[21:14:00] <cradek> this is an issue because I'm working on tool length compensation for more complex machines like 5 axis and puma
[21:14:02] <anonimasu> cradek: that makes for some interesting crashes
[21:14:15] <cradek> when you change tool length, the tool may not be pointing along Z at all
[21:16:52] <cradek> anonimasu: it's hard to know what to do to avoid crashes. Leaving the tool too low after you ask for compensation that's supposed to move it up could also be bad.
[21:18:21] <rayh> Speaking of z crashes I see that the scara defines z+ as down toward the work.
[21:18:41] <anonimasu> cradek: well, compensation alone dosent position the tool anywhere
[21:18:53] <anonimasu> you arent in contact with the work on a toolchange..
[21:19:08] <anonimasu> and most cam programs would have a z move to the clearance plane
[21:19:11] <anonimasu> and then to the work plane
[21:19:16] <cradek> rayh: hmm
[21:19:18] <anonimasu> and then slowly into the work
[21:19:27] <anonimasu> that makes sure that offsets take effect(are applied)=
[21:19:43] <cradek> anonimasu: I know - but the oddball border cases are important too
[21:20:10] <anonimasu> cradek: yeah, but you dont change tool while in cut
[21:20:27] <cradek> brb
[21:21:10] <anonimasu> cradek: sorry to butt in, but I think one move in one plane shouldnt move another axis..
[21:21:31] <anonimasu> cradek: like if you do a g0 x1.. and your tool offset if negative and your tool ramps down into your workpiece..
[21:21:54] <anonimasu> when you actually wanted to position your workpiece right below the tool
[21:22:04] <anonimasu> then do a z move down to the correct height
[21:27:23] <rayh> I can see how with a hexapod or puma like machine the RPW will change the angle at which the tool attacks the work.
[21:28:15] <rayh> With something like that the length offset should be along the tool's axis rather than a pure z offset.
[21:31:31] <anonimasu> * anonimasu nods
[21:40:10] <cradek> yes that's the idea. but, what move applies it now?
[22:03:57] <toastydeath> it would offset in all axes
[22:04:30] <toastydeath> 90 degree cutter angle to work would only have a Z component
[22:05:05] <toastydeath> 45 would have len*.707 in either X or Y, and len*.707 in Z
[23:39:42] <Ziegler> heh>> classic: http://bash.org/?244321
[23:41:53] <JymmmEMC> =)
[23:43:42] <Ziegler> we tried it on another chan I mod... never works
[23:44:01] <Ziegler> the newbs on irc just arnt newb enough
[23:46:06] <JymmmEMC> heh
[23:49:00] <cradek> well I'm sure I've talked to him a few times.
[23:49:34] <JymmmEMC> hey chris!
[23:49:45] <cradek> uh-oh
[23:50:01] <JymmmEMC> (no pretense, just a greeitng)
[23:50:20] <cradek> hi then!
[23:50:34] <JymmmEMC> like Hi Jack, but without all the ppl hitting the floors
[23:50:47] <cradek> I don't get some of these on bash.org
[23:50:53] <JymmmEMC> such as?
[23:51:08] <cradek> http://bash.org/?4618
[23:52:12] <JymmmEMC> jr highschool puberty
[23:52:38] <JymmmEMC> "sex on the brain"
[23:53:47] <cradek> this one I get: http://bash.org/?61129
[23:55:01] <cradek> and this one: http://bash.org/?70296 (substitute EMC for GTK+, [any other controller] for QT)