#emc-devel | Logs for 2007-10-14

[00:01:11] <cradek> jmkasunich__: thanks for this parport breakout thingy
[00:01:36] <jmkasunich__> you're welcome
[00:02:13] <cradek> when you gave it to me, I thought "whatever, I don't need that right this second"
[00:02:16] <cradek> but, I've used it a lot
[00:02:23] <cradek> I'm often shortsighted that way
[00:02:47] <cradek> (like the ballscrews at boeing surplus)
[00:03:03] <SWPadnos> O_O
[00:03:08] <jmkasunich__> there's a boeing surplus in wicheta?
[00:03:13] <cradek> they were $5 each
[00:03:18] <SWPadnos> O_O O_O
[00:03:18] <cradek> and I didn't take them
[00:03:21] <SWPadnos> O_O O_O O_O
[00:03:27] <jmkasunich__> !!
[00:03:54] <cradek> I just couldn't come up with any use for them right then... (um, still can't)
[00:03:55] <SWPadnos> I guess I need to drive there, maybe with a trailer
[00:04:02] <SWPadnos> CNC bandsaw!
[00:04:10] <fenn> hexapod!
[00:04:11] <cradek> there are few gems there, lots of junk you don't care about
[00:04:13] <SWPadnos> CNC - err - garage door opener!
[00:04:17] <jmkasunich__> what size?
[00:04:19] <cradek> haha
[00:04:27] <SWPadnos> a 6x12 trailer
[00:04:30] <cradek> they were shortish
[00:04:30] <SWPadnos> ;)
[00:04:49] <cradek> 16" long maybe
[00:04:58] <cradek> 12" travel maybe
[00:04:57] <jmkasunich__> long enough for a shoptask Y
[00:05:06] <SWPadnos> that's plenty for a bridgeport Z also
[00:05:30] <cradek> true but who needs that? bridgeports come with ballscrews already on them unless you really mess up :-)
[00:05:37] <SWPadnos> or get a manual mill ...
[00:05:49] <cradek> anyway...
[00:05:56] <fenn> or are currently building a hexapod with 18" screws
[00:05:56] <cradek> they're probably not there now.
[00:07:15] <SWPadnos> that's why you have to snap them up when you see them!
[00:07:26] <SWPadnos> there's also ebay, for those so inclined
[00:11:00] <jepler> cradek: maybe the RC filters mean you have much higher step lengths needed?
[00:11:07] <cradek> yep
[00:11:34] <cradek> the datasheet matches your numbers, but my drives want more
[00:11:41] <cradek> I set step reset time to 6us and it works now
[00:11:54] <SWPadnos> ouch. what's the BASE_PERIOD?
[00:11:55] <cradek> including the rotary
[00:12:28] <cradek> 100,000
[00:12:36] <SWPadnos> oh, that's quite large then
[00:12:37] <jepler> * jepler tries to remember if any of his systems *require* probe_parport
[00:19:58] <cradek> jepler: I don't spot the pin 14 problem right off
[00:22:27] <cradek> oh! yes I do.
[00:30:54] <cradek> jepler: the axis test in stepgen doesn't work for me, but the generated configuration does now
[00:34:49] <cradek> jepler: to get the rotary to work right, I set [TRAJ]MAX_VELOCITY to 999999
[00:35:04] <cradek> jepler: that needs to be higher than the highest velocity
[00:35:58] <jepler> "A" axis test?
[00:36:00] <jepler> or any axis test?
[00:36:03] <cradek> any
[00:36:19] <cradek> maybe they don't set the stepper params?
[00:37:06] <jepler> I think it misses parport.0.reset-time in the test?
[00:37:21] <cradek> that would do it
[00:39:38] <jepler> try this? http://pastebin.ca/735852
[00:39:45] <cradek> jepler: the rotary ferror are too small
[00:41:00] <cradek> jepler: I changed the default traj maxvel, so just don't write that out
[00:44:09] <cradek> jepler: that patch fixes the online test
[00:44:24] <cradek> jepler: I can set accel=0 in the online test, I don't know if you intend that or not
[00:44:59] <jepler> what nonzero accel limit should I have?
[00:45:04] <cradek> I don't know
[00:45:32] <cradek> I think you know this already - the jog buttons and the back-and-forth test have very different velocities
[00:46:22] <jepler> yeah we talked about that
[00:46:54] <SWPadnos> cradek, I think the default TRAJ limit needs to be low
[00:46:57] <cradek> the default limits on A should be -large to +large
[00:47:05] <cradek> SWPadnos: why?
[00:47:14] <Roguish_> good evening. help with a small error? 'emc/task/emctask.cc 312: interp_error: Coordinate system index parameter 5220 out of range'
[00:47:18] <SWPadnos> if a limit isn't specified in an axis section, that axis gets set to traj default
[00:47:32] <cradek> SWPadnos: oh, ouch
[00:47:40] <SWPadnos> I think it's meant to be a safe default, not a sane one :)
[00:47:49] <cradek> SWPadnos: maybe make those default to 1 instead of this one
[00:48:05] <cradek> Roguish_: your var file is messed up, edit it and make sure 5220 is set to 1.0
[00:48:14] <Roguish_> ok
[00:48:17] <cradek> Roguish_: what is the last thing you did that might have caused it?
[00:48:19] <SWPadnos> I hope nobody relies on it, but you could set TRAJ to 3, and remove the limit from all the axes, and use that one spot to change them all
[00:48:52] <Roguish_> i'm adding gantrykins to a m5i20
[00:52:15] <jepler> Roguish_: could it be this? http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingConfigurationsForDevelopmentVersions#var_file_changes
[00:55:29] <Roguish_> cradek: better. thanks. at least the axis gui comes up. where would i find a correct 'new' var file?
[00:55:45] <SWPadnos> if you changed 5221 to 1, then you now have a correct var file
[00:55:51] <SWPadnos> err - 5220
[00:56:11] <jepler> Roguish_: in the configs/common directory, for one
[00:56:49] <Roguish_> i changed 5220 to 1.
[00:57:24] <cradek> I suggest getting a new var file
[00:57:47] <cradek> to make sure it has all the right lines in it
[00:59:17] <Roguish_> just did. thanks again.
[01:10:12] <DanielFalck> jepler: thanks, I got jdi.py going. Fixed my install mess and things work.
[01:10:43] <DanielFalck> I didn't have the concept of 'run-in-place' : )
[01:11:42] <Roguish_> thanks again. i'll be back at the gantrykins in the morning.
[01:50:42] <jepler> cradek: did you or did you not spot the cause of the pin14 problem?
[01:51:22] <SWPadnos> it looked to me as though he did
[01:51:29] <SWPadnos> there was a commit mentioning a copy/paste error
[01:51:38] <jepler> oh now I see the checkin
[01:52:15] <jepler> cradek: thanks
[03:47:53] <cradek> sure
[03:52:53] <cradek> SWPadnos: did you see code to make TRAJ MAXVEL be the default for AXIS MAXVEL or were you just remembering that behavior?
[03:53:04] <SWPadnos> remembering
[03:53:06] <cradek> because I see code that defaults it to 1
[03:53:09] <SWPadnos> heh
[03:53:23] <cradek> iniaxis.cc:218
[03:53:29] <SWPadnos> I thought it used a default value for TRAJ, then used the traj value for any axis that didn't have its own setting
[03:53:41] <SWPadnos> default if there's no tRAJ setting, that is
[03:53:44] <cradek> that doesn't appear to be the case
[03:53:54] <SWPadnos> ok
[03:54:14] <cradek> 1 unit/s is a strange default, but whatever
[03:54:30] <cradek> that's actually pretty fast for inches
[03:55:25] <cradek> unless you object I'm going to leave my change for now
[03:55:54] <SWPadnos> try taking out the TRAJ MAXVEL setting and see what happens
[03:56:06] <SWPadnos> well, we know what happens I guess
[03:56:07] <cradek> I did, it does what I want
[04:01:30] <cradek> testing some more...
[04:04:57] <SWPadnos> what happens if you leave out individual axis maxvel settings?
[04:05:26] <cradek> you do get 1.0
[04:05:26] <SWPadnos> that was what I thought defaulted to [traj]maxvel
[04:05:28] <SWPadnos> ok
[04:05:34] <SWPadnos> then I have no problem leaving it as is
[04:22:48] <SWPadnos> axis, keystick, and xemc all read [TRAJ]MAX_VELOCITY
[04:22:53] <SWPadnos> I'm not sure what they do with it
[04:23:04] <cradek> they incorrectly set the top of the jog speed slider
[04:23:10] <SWPadnos> that could be a problem ;)
[04:23:43] <SWPadnos> there's got to be an NML message (or something in status) to get that setting with
[04:23:50] <cradek> axis knows how to read other things instead: MAX_LINEAR_VELOCITY, MAX_ANGULAR_VELOCITY
[04:24:17] <cradek> I don't know what the general solution is
[04:24:31] <cradek> I didn't touch any other configs; there's no harm in leaving that value in there
[04:24:41] <cradek> (except it bites people when they try to add a rotary axis)
[12:25:33] <fenn> i have what appeared 3 hours ago to be a simple task: return the x,y,z coordinates of an object in vismach
[12:26:39] <fenn> is there something missing? like reverse traversal of the tree?
[12:27:48] <fenn> i'm trying to implement a Track transform that points one object at another
[13:30:49] <jepler> fenn: you can use minigl.glGetDoublev(minigl.GL_MODELVIEW) to get the 4x4 matrix that represents the current transformation
[13:32:36] <jepler> (returned as a 16-element list, not a matrix)
[13:33:51] <fenn> yes but how do i get to the object to do that and then return it?
[13:35:08] <fenn> it seems i'd have to traverse the tree until i find the one i'm looking for
[13:38:04] <jepler> fenn: a Capture object grabs a copy of its modelview matrix each time it's traversed
[13:39:25] <jepler> .. in its self.t property
[13:39:44] <jepler> you can arrange to have a Capture at one point in the hierarchy, and use its .t at another point in the hierarchy
[13:39:51] <fenn> so i can do object.t.capture()?
[13:40:08] <fenn> or is t the matrix i want?
[13:40:20] <jepler> Capture is an object you put in the tree, like a Sphere or a Collection
[13:40:33] <fenn> ok
[13:40:41] <jepler> its .t gets updated when that part of the tree is traversed
[13:41:08] <fenn> * fenn tries
[13:41:10] <jepler> (by the automatic redraw procedure)
[13:46:06] <jepler> # note that this tranforms from the current coordinate system
[13:46:06] <jepler> # to the viewport system, NOT to the world system
[13:46:51] <fenn> cant you get the world system with a different gl call?
[13:47:42] <jepler> no, the world system isn't really stored anywhere by opengl
[13:48:23] <fenn> so i will need to pass the tool Capture to my function also right?
[13:49:13] <fenn> where tool is the tool coordinate system
[13:49:44] <fenn> or i guess i could just do Capture() and pass that
[13:50:33] <fenn> and then use your view2world
[13:50:43] <jepler> you want to have a reference to the Capture() object, and look at its .t
[13:51:27] <fenn> and that's all i need to do to get world coordinates?
[13:51:28] <jepler> you'll probably need a second Capture() object which is at the outermost level of the model; its .t will reflect the transformation from model to view before any other transformations are applied
[13:52:14] <jepler> read the gymnastics that vismach.py:O.redraw() does to make the 'backplot'
[13:52:24] <jepler> there are a lot of steps and I'm having trouble articulating them ..
[13:53:47] <jepler> not also the comment on the invert() function that it does not work for matrices which include scaling; I think you added scaling classes?
[13:54:13] <fenn> yeah i noticed that.. i might just remove the scaling stuff
[13:55:49] <fenn> is there a better way to revert a cvs change than: check out old revision, copy to another name, delete old revision, check out new revision with no tags, copy old file to original name, commit
[13:58:19] <cradek> cvs up -p -r1.2.3.4 file > file
[13:59:19] <fenn> thanks
[17:23:53] <Roguish_> Roguish_ is now known as Roguish
[23:08:46] <fenn> stepper-gantry/stepper.hal:22: setp requires 2 arguments, 3 given
[23:09:27] <fenn> easily fixed, but i wonder when/why that happened
[23:09:53] <fenn> its INPUT_SCALE = 150.333 0
[23:15:17] <fenn> nevermind, i was just out of date
[23:54:12] <jmkasunich__> jmkasunich__ is now known as jmkasunich