#emc | Logs for 2007-01-15

Back
[00:01:15] <CIA-8> 03cradek 07HEAD * 10emc2/configs/nist-lathe/README: description
[00:02:09] <cradek> alex_joni: do I now have to cvs add in the branch checkout too?
[00:03:37] <A-L-P-H-A> alex_joni, the benevolent dictator of the CVS. :)
[00:03:59] <dave_1> well, beings that it is quiet ... what are my options for running steppers with encoders under emc?
[00:04:05] <cradek> nah, that's me, I was just hoping he knows how to use it :-)
[00:04:24] <cradek> dave_1: what's your goal in using encoders?
[00:04:41] <dave_1> paranoid!
[00:04:48] <alex_joni> cradek: yes
[00:05:17] <CIA-8> 03cradek 07v2_1_branch * 10emc2/configs/nist-lathe/README: description
[00:05:19] <cradek> aha
[00:05:20] <alex_joni> A-L-P-H-A: I just did that for a couple of files.. so I happen to know
[00:05:39] <alex_joni> don't forget the pluto-lathe :D
[00:05:51] <dave_1> Actually I want to cheat a bit on emc.... I need to drive 2 axes in coordinate motion and feedhold rather often
[00:06:02] <alex_joni> dave_1: ok..
[00:06:11] <dave_1> One axis relative to another would do.
[00:06:26] <alex_joni> how do you mean?
[00:06:38] <dave_1> where one axis is driven mechanically ....
[00:06:54] <dave_1> and the other move relative to it.
[00:07:07] <cradek> interesting
[00:07:14] <dave_1> Do you know what a shaper is... that is a metal shaper.
[00:07:22] <cradek> sure
[00:08:23] <dave_1> OK ... so one axis can run off the rachet, the ram will take care of itself and I need to move Z in relation to the racheted axis
[00:08:42] <cradek> Z?
[00:08:50] <dave_1> the tool
[00:09:07] <cradek> maybe I don't know what a shaper is then, I thought they just cut flats
[00:09:15] <cradek> you can change height?
[00:09:16] <dave_1> well pretty much
[00:09:29] <dave_1> height is now manual
[00:09:42] <cradek> ok
[00:10:30] <dave_1> It might get a little interesting ... driving the z which is moving back and forth at once to twice a sec.
[00:10:43] <dave_1> ie z is on the ram
[00:10:45] <cradek> I'm struggling to see how you can use gcode for this ... I'm not sure how you would describe the shape you want
[00:10:56] <dave_1> plain old gcode
[00:11:31] <dave_1> if the ram moves in Y and the crossfeed is X then you adjust z to cut a profile
[00:12:46] <cradek> but X and Y both are not controlled axes right?
[00:13:12] <cradek> I mean doesn't Y move on its own and X is ratcheted by Y?
[00:13:17] <A-L-P-H-A> http://bash.org/?82127
[00:13:22] <dave_1> right only X and Z or Z relative to X
[00:17:35] <CIA-8> 03cradek 07HEAD * 10emc2/configs/lathe-pluto/README: description
[00:17:36] <lerneaen_hydra> A-L-P-H-A: funnily enough, I liked requiem for a dream more than pi
[00:18:10] <CIA-8> 03cradek 07v2_1_branch * 10emc2/configs/lathe-pluto/README: description
[00:22:31] <A-L-P-H-A> never saw RFAD.
[00:24:09] <dave_1> cradek just sent you a screenshot, maybe that will help
[00:24:46] <lerneaen_hydra> A-L-P-H-A: oh? see it
[00:24:50] <lerneaen_hydra> real good
[00:26:44] <dave_1> btw -- is elsons site down?
[00:26:49] <dave_1> brb
[00:27:57] <alex_joni> dave_1: seems so here
[00:28:18] <dave_1> tnx
[00:30:22] <alex_joni> good night all
[00:30:31] <dave_1> bye alex
[00:31:27] <dave_1> be back in a while
[00:35:45] <A-L-P-H-A> lerneaen_hydra, will do... I'll look for it next time.
[00:36:32] <lerneaen_hydra> great :)
[00:38:13] <robin_sz> alex_joni, dont forget to send me some free ISO9409 T50 flanges, ideally with weld torch holders on them
[00:38:15] <robin_sz> ;)
[00:39:47] <lerneaen_hydra> 'night
[00:52:28] <CIA-8> 03jmkasunich 07HEAD * 10emc2/src/hal/ (hal_lib.c hal_priv.h): track OUT and IO pins connected to a signal separately, and enforce consistent rules for connecting such pins to signals, also make link() properly handle links that already exist
[00:52:49] <CIA-8> 03jmkasunich 07HEAD * 10emc2/src/hal/utils/halcmd.c: add some comments, modify net command to handle all combinations of OUT and I/O pins
[01:10:28] <dave_1> OK, got stuff cooking and I'm back for a bit
[01:15:40] <skunkworks> dave_1: we have a shaper. The tabe moves left and right - up and down - was thinking of puting a ball screw on the head (in/out) and have a very solid xyz machine. (mounting a spindle on the ram)
[01:17:32] <dave_1> skunkworks: well, I'm going the other way.
[01:17:47] <dave_1> just be able to profile with it
[01:18:06] <dave_1> encoders on cross-feed and Z (ram)
[01:18:52] <jmkasunich> so crossfeed will move at a constant rate?
[01:19:07] <jmkasunich> you aren't gonna be able to do vertical (or even steep) surfaces that way
[01:19:36] <dave_1> some people have used them as surface grinders ... mount a circular table on the table (rotational) and a grinding wheel on the ram and recip the ram slowly in and out
[01:20:32] <jmkasunich> if you control both crossfeed and table up/down and let the ram reciprocate, its sort of a flat version of a lathe
[01:20:52] <jmkasunich> and you could treat it like a lathe config, complete with the lathe tool compensation, etc
[01:21:42] <jmkasunich> in fact, you could get creative and use lathe spindle sync to do your feeding
[01:21:52] <jmkasunich> you could make if feed only when the ram is back
[01:21:53] <dave_1> jmk: for now this is a very limited need but still important.
[01:22:18] <jmkasunich> I'm just wondering how you're gonna program your part shape
[01:22:35] <jmkasunich> when the cross axis isn't controlled, and the vertical one is
[01:25:34] <dave_1> jmk: Oh! both X (cross and Z downfeed) need to be related ... Y is ram (tool)
[01:25:47] <dave_1> just sent you a png. :-)
[01:26:27] <dave_1> so either we drive both X and Z or we make Z related to the position of X
[01:26:34] <jmkasunich> thats the part you want to make?
[01:26:40] <dave_1> yes
[01:27:00] <dave_1> I can cnc most of it and then clean up the radii on the shaper
[01:27:10] <jmkasunich> are you gonna put it on some kind of rotary, or do it in a couple passes, flipping between
[01:27:25] <jmkasunich> oh, you don't need to shape the whole thing
[01:28:09] <jmkasunich> so which radii are you cleaning up
[01:28:11] <dave_1> exactly. ... takes too long to do the whole thing on the shaper... just the cleanup of the sharp corners on the right of the piece
[01:28:46] <jmkasunich> how will the part be oriented when shaping? right end facing up?
[01:28:49] <dave_1> the rest goes well with a A axis
[01:29:09] <jmkasunich> or as shown in the drawing?
[01:29:31] <dave_1> I'll have to flip it using the right hand hole and a pin as references
[01:29:54] <jmkasunich> duh, you are talking about cleaning up the sharp inside corners, right?
[01:30:10] <jmkasunich> so two corners on the top and one on the bottom
[01:30:18] <dave_1> yes
[01:30:32] <jmkasunich> only this part?
[01:30:41] <jmkasunich> IOW, its pretty much a dedicated machine?
[01:30:52] <dave_1> for now exactly
[01:31:21] <jmkasunich> the problem is that g-code describes a toolpath with respect to time (feedrate, etc)
[01:31:46] <jmkasunich> and you want a profile of Z vs X
[01:31:56] <dave_1> YEP
[01:32:36] <jmkasunich> if its truly dedicated, you can use some other approaches
[01:32:47] <dave_1> lacking a 5 axis machine this is difficult otherwise. I can still see the shaper marks on the original part
[01:33:01] <dave_1> like?
[01:33:14] <jmkasunich> could you write a C function that would take the X position as an input and return the Z position?
[01:33:32] <dave_1> OK ... to drive hal?
[01:33:35] <jmkasunich> IOW, you'd need to do the math to figure the intersection of the curve and straight parts
[01:33:36] <jmkasunich> yeah
[01:33:43] <dave_1> hmmm
[01:34:13] <jmkasunich> encoder generates X, goes thru a custom module with your function in it, Z comes out and goes to stepgen or PID or whatever
[01:34:26] <dave_1> interesting idea ...
[01:34:34] <jmkasunich> like an electronic cam
[01:34:48] <jmkasunich> which gives me an even screwier idea
[01:34:53] <dave_1> I've got to go cook ... so I'll get back to you later.... this has posibilities
[01:34:59] <jmkasunich> why not use one part as a cam
[01:35:03] <jmkasunich> think tracer mill
[01:35:30] <dave_1> I've thought about that... z is pretty stiff. ....
[01:35:53] <jmkasunich> mount two parts on the machine, one inline with the ram, the other offset
[01:35:52] <dave_1> takes significant torque to drive the ball screw
[01:36:09] <jmkasunich> have a finger that rides on the pattern
[01:36:26] <jmkasunich> if the finger drops, you run Z up (electrically)
[01:36:34] <jmkasunich> if the finger rises, lower Z
[01:36:41] <dave_1> let me think about that in view of the forces involved
[01:36:52] <dave_1> .... null the sensor?
[01:37:01] <dave_1> as in a real follower
[01:37:09] <dave_1> gotta go.
[01:37:20] <jmkasunich> what you need is a finger that can indicate "too high" "too low" or "just right" with a range of 0.001 or so, and an electrical output
[01:37:32] <jmkasunich> send that to a parport input and use HAL to close the loop
[01:41:59] <Jymmmm> .001 is kinda tight tolerance, isn't it?
[01:42:21] <Jymmmm> for an electrical "switch"
[01:42:41] <jmkasunich> yeah
[01:42:53] <jmkasunich> the real requirement depends on the specs for the parts he's making
[01:43:10] <Jymmmm> I have a .003 switch, but it's $40 USD too
[01:43:40] <jmkasunich> this guy is doing production work - if it solved his problem, a $40 switch is cheap
[01:44:14] <Jymmmm> true, but I think a 10 turn linear pot might do better
[01:44:56] <jmkasunich> pot might be a good idea
[01:45:01] <jmkasunich> but I'd use a one turn
[01:45:11] <jmkasunich> you want a small mechanical change to cause a large electrical one
[01:45:21] <jmkasunich> ten turn pot does the oppostite
[01:45:24] <jmkasunich> oppsoite
[01:45:29] <jmkasunich> opposite
[01:45:32] <jmkasunich> heh
[01:45:44] <Jymmmm> I only mention the 10 turn, becasue I've always foudn them to be made a LOT better and smoother than the typical $0.39 ones
[01:46:17] <Jymmmm> and if you're mounting it on the slide... you ahve the extra movement
[01:57:44] <CIA-8> 03jmkasunich 07HEAD * 10emc2/src/hal/utils/halcmd.c: allow net command even if all pins are already connected to the signal
[02:03:42] <CIA-8> 03jmkasunich 07HEAD * 10emc2/docs/src/hal/rtcomps.lyx: document change from params to pins for siggen
[02:07:29] <ds3> is there a target in the source to put the entire tree back to checked out state, like make distclean or something similar?
[02:08:26] <cradek> make clean is something like that
[02:08:32] <cradek> also cvs up -C is sometimes useful
[02:10:29] <ds3> hmmmm
[02:10:44] <ds3> still not doing the same thing as when I checked it out
[02:11:43] <ds3> can anyone confirm these instructions are still good: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?EMC2_Pure_Simulator
[02:12:07] <cradek> yes that looks right
[02:12:22] <cradek> say more specifically what is wrong
[02:13:05] <ds3> it all started with it not finding my tcl.h file even though I gave it an explicit CPPFLAGS=-I.... and CFLAGS=-I....
[02:13:37] <cradek> configure should find it without those acrobatics
[02:13:40] <ds3> so I figure something in head is broq, a cvs update to a known label was the same; so I cvs upd -A to bring it back and it has been more broken since
[02:14:09] <ds3> I have an unusual install of tcl/tk; I am trying it make it run out of /opt/emc entirely w/o looking at the stock system stuff
[02:14:23] <cradek> checking for tcl... /usr/lib/tcl8.4/tclConfig.sh found
[02:14:23] <cradek> checking for tk... /usr/lib/tk8.4/tkConfig.sh found
[02:14:29] <ds3> this is what I used: ./configure --enable-run-in-place --enable-simulator --prefix=/opt/emc --wit
[02:14:30] <ds3> h-tclConfig=/opt/emc/lib/ --with-tkConfig=/opt/emc/lib/
[02:14:41] <ds3> I didn't install tcl/tk in /usr
[02:15:04] <ds3> let get a fresh tree and see if it is any different
[02:15:12] <cradek> it won't be
[02:16:12] <cradek> did it say it found your tclConfig.sh?
[02:16:31] <ds3> the very first time, yes. it hasn't since then. that's why I was looking for a distclean or like
[02:17:27] <cradek> it looks like --with-tclConfig= should be your full path to tclConfig.sh INCLUDING the filename
[02:17:34] <ds3> tried that
[02:17:39] <cradek> maybe you did it differently the first time if it worked?
[02:18:23] <ds3> Let me start with a clean tree as I had to do other acrobatics to figure out what was wrong (i.e. remove '@'s in the makefile)
[02:18:41] <cradek> oh, ok, I didn't realize you changed files
[02:18:55] <ds3> those _SHOULD_ be harmless changes
[02:19:02] <cradek> cvs up -dPAC will put them back
[02:19:19] <ds3> all I did was make it show the gcc/g++ so I see other oddness
[02:19:59] <cradek> oh ok, the @ that means "don't print this", not the @VAR@ stuff
[02:20:28] <ds3> yes. I am on a x86_64 system and sometimes stock gcc/g++ commandlines don't find the right versions
[02:20:40] <cradek> I see
[02:21:08] <cradek> we have one developer who is running emc2/simulator on a 64 system. you can build it as 64.
[02:21:58] <CIA-8> 03jmkasunich 07v2_1_branch * 10emc2/docs/src/hal/rtcomps.lyx: backport siggen documentation
[02:22:02] <ds3> I am trying to build it as 64 but it takes some shanagan with the way the libreadline dependancies are setup; 64bit files are in /usr/lib64 but it tries to use /usr/lib...but that's minor
[02:24:54] <cradek> if you're good at fixing up any problems with configure, patches happily accepted :-)
[02:25:11] <ds3> is this a manually written configure?
[02:25:34] <ds3> I'll be happy to work out patches, etc once I get a working binary =)
[02:25:47] <cradek> configure.in is manually written, yes
[02:25:56] <ds3> wtf, I did exactly as I described and it works as well as it did the first time
[02:26:13] <ds3> now I need to patch the Makefile to find libreadline correctly :/
[02:27:35] <cradek> libreadline is only used in halcmd, and it works without too...
[02:28:35] <ds3> the makefile is a bit odd in how it does makes /usr/lib/libreadline.so a dependancy
[02:28:46] <ds3> okay getting much much further
[02:29:05] <ds3> this is going messy
[02:29:52] <ds3> things are blowing up in hal_lib.c
[02:30:01] <ds3> :1
[02:30:04] <ds3> sorry
[02:30:05] <cradek> use pastebin.ca
[02:30:20] <cradek> I type vi commands in irc all the time too
[02:32:42] <cradek> I don't get libreadline.so as a dependency anywhere...?
[02:33:35] <ds3> -../bin/halcmd: $(call TOOBJS, $(HALCMDSRCS)) ../lib/libemc.a ../lib/libnml.so ../lib/libemchal.so $(READLINE_LIBS)
[02:34:13] <ds3> that line - $(READLINE_LIBS) is expanded and results in the @$(CXX) getting passed /usr/lib/libreadline.so explicitly cuz of the $^ expansion
[02:34:40] <cradek> ok
[02:34:42] <ds3> in my case, Ijust removed the $(READLINE_LIBS) and explicitly added -L/usr/lib64 -lreadline -ltermcap since that's what is defined as
[02:35:07] <cradek> so configure is finding the wrong readline on your system?
[02:35:39] <ds3> not sure if I can blame configure for this one. it really should not refer to it as /usr/lib/libreadline.so
[02:36:04] <cradek> well it's not
[02:36:06] <cradek> configure: READLINE_LIBS="-lreadline -ltermcap"
[02:36:18] <ds3> the traditional/"correct?" way of doing it is to let the linker find it via the appropriate config
[02:36:40] <SWPLinux> on my ubuntu/AMD64 system, /usr/lib64 is a link to /usr/lib
[02:36:42] <cradek> sure, with -l
[02:36:51] <ds3> I removed the '@' in the compile line and it showed me with the explicit refernece
[02:37:11] <ds3> SWPLinux, how does it handle mix enviroment? (32 vs 64)
[02:37:14] <cradek> in config.log what is READLINE_LIBS?
[02:37:21] <cradek> READLINE_LIBS='-lreadline'
[02:37:21] <SWPLinux> there's a lib32 dir as well
[02:37:32] <SWPLinux> so lib32 is special, and lib = lib64
[02:37:38] <ds3> READLINE_LIBS='-lreadline -ltermcap'
[02:38:00] <ds3> SWPLinux: ah, so you'll have the inverse problem if you have to build 32 bit binaries
[02:38:02] <cradek> and in Makefile.inc?
[02:38:08] <SWPLinux> ds3: I suppose so
[02:38:39] <ds3> PREADLINE_LIBS = -lreadline -ltermcap
[02:38:45] <ds3> s/P//
[02:39:17] <cradek> so it's right now, but it wasn't before?
[02:39:24] <ds3> off the top of my head, I suspect it is a make thing that expands it that way
[02:39:56] <SWPLinux> there is a --libdir option for configure
[02:40:18] <SWPLinux> but I haven't looked at whether that's input libs or output :)
[02:40:28] <cradek> ds3: I don't think it's set that way now. I don't see anything that could expand it.
[02:40:42] <ds3> bingo!
[02:40:43] <ds3> http://web.mit.edu/6.033/labdoc/make_4.html#SEC31
[02:40:59] <ds3> see that explanation. make doesn't know about /usr/lib64
[02:42:34] <cradek> huh, I get that expansion too, but I don't see why
[02:42:56] <ds3> that link explains make
[02:43:52] <cradek> No need to remake target `-lreadline'; using VPATH name `/usr/lib/libreadline.so'.
[02:44:29] <SWPLinux> can you modify the "matching vpath or VPATH" search paths using configure options?
[02:44:36] <ds3> IMO, it should not have $(READLINE_LIBS) in the dependacy line
[02:44:44] <SWPLinux> * SWPLinux is just throwing out ideas, and is not a make / configure expert
[02:44:56] <ds3> not very useful to have it there unless you expect people to be rebuilding libreadline often ;)
[02:45:10] <cradek> yes we could move it down to the CXX line
[02:45:35] <ds3> but that's, IMO, minor =)
[02:45:57] <ds3> pasting stuff together for another question... a more complex one :(
[02:47:26] <CIA-8> 03cradek 07HEAD * 10emc2/src/hal/utils/Submakefile: don't cause make to blindly expand -lreadline according to VPATH, which breaks linking according to LDFLAGS's library search path
[02:47:49] <ds3> wow, talking about quick turnaround ;)
[02:47:58] <cradek> thanks for convincing me it was a problem :-)
[02:48:29] <SWPLinux> how hard is it to add a library prefix option to configure?
[02:48:46] <cradek> are you confusing me with someone who knows configure!?
[02:48:51] <SWPLinux> I may be ;)
[02:48:59] <cradek> * cradek wonders if Unicode has an interrobang
[02:49:00] <SWPLinux> compared to me, you're the Dalai Lama
[02:49:14] <SWPLinux> (as far as configure, anyway)
[02:49:52] <ds3> it claims to support that automatically via the LDFLAGS variable i.e. LDFLAGS="-L/new/library/path" ; this is per ./configure --help
[02:49:58] <ds3> okay - http://pastebin.ca/316905
[02:50:39] <SWPLinux> this is a simulator build?
[02:50:45] <ds3> yep
[02:50:58] <cradek> you sure don't need the proc stuff then
[02:51:10] <cradek> what's on or before proc_fs.h:44?
[02:51:40] <ds3> proc_fs.h is part of the kernel but _
[02:51:43] <cradek> some other folks have had terrible times with the system includes on new 2.6 kernels
[02:51:53] <ds3> typedef int (read_proc_t)(char *page, char **start, off_t
[02:51:54] <ds3> ...
[02:52:14] <ds3> Oh boy... this is 2.6.15.5
[02:52:28] <ds3> I can comment out that entire section then
[02:53:29] <ds3> cool... build finished
[02:53:33] <cradek> woo
[02:53:39] <cradek> does it run?
[02:53:54] <SWPLinux> hmmm. shouldn't proc stuff be disabled in a simulator build? (or is it just me?)
[02:54:07] <ds3> that's the next thing to figure out how to do it... I have runinplace setup
[02:54:15] <cradek> scripts/emc
[02:54:27] <cradek> from the checkout directory
[02:54:51] <ds3> not quite
[02:55:17] <cradek> it would be no fun if it worked on the first try.
[02:55:35] <ds3> axis bombs, mini bombs
[02:55:48] <SWPLinux> you did mention a weird tcl setup :)
[02:56:00] <cradek> you're trying to run the configs under sim?
[02:56:13] <ds3> yeah... trying sim - axis, sim - tkemc, etc
[02:56:21] <cradek> ok, so the config chooser does run, that's a good sign
[02:56:30] <cradek> how about xemc?
[02:56:38] <ds3> all of them bomb out similarly
[02:56:48] <cradek> and how's that exactly
[02:57:04] <ds3> since I don't know what is significant, let me create another pastebin
[02:57:09] <cradek> ok
[02:59:11] <ds3> let me try something first
[02:59:32] <SWPLinux> you know, it takes significantly longer to load a 21M nc file in AXIS than it takes to load a 2k file :)
[02:59:41] <cradek> oh imagine that
[02:59:53] <SWPLinux> 515634 lines - that's pretty large-ish
[03:00:08] <cradek> is it still usable?
[03:00:26] <SWPLinux> I can rotate / zoom / pan in fluid realtime
[03:00:34] <cradek> good
[03:00:40] <cradek> are any of those arcs, or all g1?
[03:00:49] <SWPLinux> this is on the quad-core opteron machine wuth the 7800GT though
[03:01:02] <SWPLinux> all G1, I bet
[03:01:22] <cradek> arcs can make a lot of segments fast
[03:01:34] <SWPLinux> it's output from something like image-to-gcode
[03:02:14] <SWPLinux> you should see the number og G0 moves. I think it's set up to cut in only one direction
[03:02:15] <cradek> a g1 for each pixel, or does it merge colinear points into lines?
[03:02:18] <SWPLinux> and rapid back
[03:02:25] <SWPLinux> lines, it seems
[03:02:42] <cradek> well, that's good it's not totally stupid :-)
[03:02:44] <SWPLinux> I posted a screenshot a while ago. lemme find it for you
[03:02:45] <SWPLinux> heh
[03:02:52] <ds3> okay here it is after fixing one stupid thing - http://pastebin.ca/316912
[03:02:53] <cradek> can you upload it somewhere?
[03:03:25] <SWPLinux> it is somewhere - I'm checking where
[03:03:35] <cradek> ds3: yuck
[03:03:43] <ds3> eh?
[03:03:51] <cradek> ds3: your tool table is too big for the nml buffer
[03:04:06] <cradek> let me see if I can figure out where to change it
[03:04:09] <ds3> I see. except, AFAIK, I have no tools =)
[03:04:27] <cradek> you do... I think there's a tool table in sim
[03:04:30] <SWPLinux> http://www.cncgear.com/images/Screenshot.png
[03:04:36] <ds3> Oh
[03:04:38] <SWPLinux> it's big - that was across both screens :)
[03:05:15] <cradek> ds3: in emc/nml_intf/canon.hh reduce CANON_TOOL_MAX
[03:05:27] <SWPLinux> or try increasing the size of the buffer to 8192 ...
[03:05:47] <SWPLinux> wasn't that the 64-bit fix?
[03:05:57] <cradek> maybe, I didn't know of that
[03:06:27] <cradek> in emc.nml, line toolSts
[03:06:34] <cradek> try doubling the size
[03:06:35] <SWPLinux> right. chage it to this:
[03:06:42] <cradek> that's got to be better than my recommendation
[03:06:45] <SWPLinux> B toolSts SHMEM localhost 8192 0 0 5 16 1005 TCP=5005 xdr
[03:06:48] <ds3> building it with the smaller number of tools instead
[03:07:00] <cradek> ok, either will work I think
[03:07:06] <SWPLinux> why configure with text files when you can recompile :)
[03:07:09] <cradek> haha
[03:07:28] <cradek> wonder if we should change the default nml file
[03:07:51] <SWPLinux> nope. I think that caused problems on some systems
[03:08:02] <cradek> * cradek shivers
[03:08:03] <SWPLinux> and it dodn't fix it on all 64-bit ones either, IIRC
[03:08:06] <SWPLinux> didn't
[03:08:10] <cradek> ??
[03:08:12] <ds3> 56 is a lot of tools.. the smaller industrial machine (220V powered) don't have 56 tools!
[03:08:30] <SWPLinux> the bigger ones don't either, most of the time (at least, not at once)
[03:08:53] <ds3> Haas claims to make a machine with a huge changer
[03:08:54] <cradek> ds3: some machines have carousels that go all around the machine... but nobody has that many holders I bet!
[03:09:10] <SWPLinux> only people with those machines ;)
[03:09:22] <ds3> not saying they don't exists, just saying it'd be unusual for most EMC users
[03:09:22] <cradek> wonder how many skunkworks's machine has - seems like it looked really big
[03:09:35] <cradek> sure
[03:10:33] <SWPLinux> man. I really wish I had the time and inclination to make RTAI work on AMD64/SMP
[03:10:41] <ds3> think this problem is mine
[03:10:50] <cradek> still not working?
[03:10:56] <SWPLinux> well, I have the inclination to get it to work, but not the inclination to spend that amount of time on it
[03:11:28] <ds3> tk problems this time
[03:11:31] <cradek> SWPLinux: seems the AXIs preview is not very useful for that file...
[03:11:50] <SWPLinux> nope
[03:12:15] <SWPLinux> but it does bring up an interesting new feature idea - make it possible to turn off certain kinds of moves
[03:12:15] <cradek> that sure does look like a lot of useless G0 moves
[03:12:22] <SWPLinux> (which can also be done if the colors are user-definable)
[03:12:42] <ds3> sigh. python/tk issues ack... (got to get tkinter setup)
[03:12:57] <cradek> short of turning them all off you're not going to get a better view of that file
[03:13:14] <cradek> ds3: but can you run tkemc or xemc?
[03:13:19] <SWPLinux> I think it would look OK witjout the G0 moves obscuring the G1's
[03:13:29] <ds3> good point!
[03:14:42] <cradek> SWPLinux: if your gcode generator was smarter it could cut the file size by 3/4
[03:14:50] <ds3> cool.... MDI works w/tkemc
[03:15:00] <cradek> SWPLinux: there's lots of redundant stuff on every line, and line numbers are just stupid
[03:15:01] <SWPLinux> it's not my generator ;) (or my daughter in the photo)
[03:15:06] <cradek> ds3: great
[03:15:12] <ds3> thanks!
[03:15:25] <cradek> very welcome
[03:15:40] <SWPLinux> ds3: the preview feature would be *much* better with AXIS - you may still want to set up tkinter ...
[03:15:43] <cradek> and we got one bugfix for the next guy to enjoy
[03:15:59] <ds3> SWPLinux: I intend to
[03:16:00] <cradek> yeah, for simulating/preview/developing AXIS is priceless
[03:16:23] <ds3> goal here is to avoid buying a copy of cutviewer for $300 ;)
[03:17:06] <SWPLinux> note that all the existing graphical preview functions in emc are for the toolpath, not the material
[03:17:15] <SWPLinux> ie, you'll see where the tool goes, but not what the resulting part looks like
[03:17:40] <SWPLinux> patches gratefully accepted ;)
[03:17:48] <cradek> heh that's my line
[03:17:52] <SWPLinux> sorry ;)
[03:18:08] <ds3> I just want to know if that program will result in me buying a new table or spindle ;)
[03:18:43] <SWPLinux> heck, just get an all-new machine
[03:18:45] <SWPLinux> a really big one
[03:20:50] <SWPLinux> hmmm. I thought gcc-doc provided manpages for C library functions, but it doesn't seem to
[03:20:57] <cradek> a few holes in the table never hurt anyone
[03:20:58] <SWPLinux> does anyone know what package gives that?
[03:21:30] <cradek> apt-file search printf.3
[03:21:35] <ds3> stupid question, how do I 'reset' the machine (restart my G code program) in mini?
[03:21:42] <cradek> * cradek teaches a man to fish ... for manpages
[03:21:54] <cradek> ds3: just hit run again?
[03:22:06] <ds3> n/m
[03:22:09] <cradek> hit escape to abort, hit run again
[03:22:20] <ds3> found out emc does that.... used to the HAAS where there is a reset button
[03:22:20] <SWPLinux> program menu -> reset
[03:22:22] <ds3> sorry
[03:25:11] <SWPLinux> yay - glibc-doc
[03:25:52] <cradek> manpages-dev: usr/share/man/man3/snprintf.3.gz
[03:25:56] <cradek> hmm
[03:26:02] <cradek> for me it's this
[03:26:39] <SWPLinux> argh - that wasn't it
[03:26:43] <cradek> ^^
[03:26:53] <SWPLinux> right :)
[03:27:26] <SWPLinux> heh - here's one we all need: paul-doc ;)
[03:27:48] <cradek> Description: Documentation for paul in German language
[03:27:50] <cradek> haha
[03:27:54] <SWPLinux> yeah :)
[03:54:49] <ds3> anyone tried axis with Python 2.5?
[03:55:20] <cradek> I think I remember it doesn't quite work
[03:55:31] <cradek> something about the C interface changed ... or something
[03:55:37] <cradek> jepler would know
[03:55:46] <ds3> thank you. *grumble*
[03:56:51] <cradek> is it a compile error in halmodule.cc?
[03:56:55] <ds3> yep
[03:56:59] <cradek> invalid conversion something?
[03:57:02] <ds3> hal/halmodule.cc:437: error: invalid conversion from 'int (*)(PyObject*)' to 'Py_ssize_t (*)(PyObject*)'
[03:57:15] <ds3> grabbing 2.4 and building it now
[03:57:16] <cradek> yep
[03:57:44] <cradek> I hope that's a 2.5 problem and not a 64 bit problem... I'm not sure
[03:58:06] <cradek> we had an unhelpful bug report once before but no details
[03:58:29] <ds3> once I get it running, I'll be making 32 bit bins
[03:59:06] <ds3> and if all goes well, I just tar off /opt/emc and I have a semiportable tar ball
[03:59:49] <cradek> http://www.mail-archive.com/[email protected]/msg00229.html
[04:00:14] <cradek> "This is a fallout from the move to python 2.5, hope this will be fixed soon,"
[04:01:04] <cradek> looks like everyone has this problem with python2.5 and c++
[04:01:46] <cradek> http://mail.python.org/pipermail/python-dev/2006-March/062561.html
[04:02:07] <cradek> looks like this preprocessor trick is the "right" fix
[04:02:44] <cradek> patches gratefully accepted :-)
[04:03:21] <ds3> sounds like a 64bit problem
[04:03:29] <ds3> rebuilding with 2.4 instead ;)
[04:03:41] <cradek> ok
[04:03:48] <cradek> we'll get to the bottom of it someday
[04:05:22] <ds3> wheeeeeeeeeee 2.4 makes axis run
[04:05:30] <cradek> yay!
[04:07:30] <ds3> this really beats the simulation mode on a HAAS by a million miles
[04:08:01] <cradek> :-)
[04:08:04] <cradek> that makes my day
[04:08:29] <ds3> have you used the HAAS simulators?
[04:08:38] <cradek> no I've never used a machine with a commercial controller
[04:09:08] <cradek> sometimes it's good, sometimes bad, to not know how the other guys "do it"
[04:09:11] <ds3> well, they are 2D only for one and the Haas is the friendiest I have seen... the Fanuc is worse
[04:09:51] <cradek> jeff and I are really proud of the navigation and interactivity in AXIS (click one of the lines in the preview, or a line of gcode)
[04:10:41] <SWPLinux> was there ever a consensus about adding an edit button? (to run an external editor)
[04:11:22] <cradek> SWPLinux: consensus and submitted patch, no...
[04:11:26] <SWPLinux> heh
[04:11:42] <SWPLinux> I recall some resistance to the idea, and wasn't sure if a patch would be "gratefully accepted" :)
[04:12:17] <cradek> I read somewhere that groups have changed from saying "patches gratefully accepted" to "patches thoughtfully considered"
[04:12:24] <SWPLinux> heh
[04:12:32] <cradek> I'm careful to always use the right one
[04:12:38] <ds3> hmmm the dialect is definitely not haas like (gcode dialect)
[04:12:46] <SWPLinux> patches (conforming to our notion of good coding style nad containing functions we like) possibly accepted
[04:13:19] <Jymmm> * Jymmm rolls eyes at SWPLinux
[04:13:25] <cradek> ds3: huge differences or just a few things?
[04:14:08] <ds3> cradek: somewhat large differences it seems. so far - O codes are not supported (should be easy to fix, make that a no-op) and R planes are not modal
[04:14:32] <SWPLinux> O codes should be supported, but the syntax may not be the same
[04:14:37] <cradek> we have O codes but they're probably totally different (looping/subroutines)
[04:14:44] <cradek> I don't know what R planes are?
[04:14:50] <ds3> it barfs on O codes
[04:15:15] <ds3> R planes are used in the drill/tap/bore can cycles; specifies how far the tool should retract; the 'Retract plane'
[04:15:29] <cradek> ah
[04:15:40] <ds3> on HAAS and I think FADAL, R is model so it does not need to be repeated unless you change it. EMC faults if R is not specified
[04:16:21] <SWPLinux> I assume it gets reset at program end or something?
[04:16:33] <ds3> Reset? what do you mean?
[04:16:48] <cradek> For canned cycles, we will call a number "sticky" if, when the same cycle is used on several lines of code in a row, the number must be used the first time, but is optional on the rest of the lines. Sticky numbers keep their value on the rest of the lines if they are not explicitly programmed to be different. The R number is always sticky.
[04:16:50] <SWPLinux> ie, it should be an error if no R code has been encountered in a particular program
[04:17:09] <cradek> the spec says in G81..G89 R is "sticky"
[04:17:13] <cradek> is that not true?
[04:17:22] <cradek> which cycle is the problem?
[04:17:23] <ds3> that is not the behavior I am seeing
[04:17:28] <SWPLinux> ok - that's the behavior I was thinking of
[04:17:43] <ds3> the first instance of G81 I have:
[04:17:44] <ds3> G90 G81 G99 Z-0.0023 R0.010 F10.0^M
[04:18:04] <ds3> then it faults on this line with a complain about missing R plane:
[04:18:04] <ds3> G90 G81 G99 Z-0.0008 ^M
[04:18:30] <ds3> sorry about the ^M; seems HAAS machines want that. I'm testing my old programs that ran on Haas
[04:19:45] <cradek> a program made with those two lines loads and runs for me
[04:20:10] <ds3> maybe it is axis? I get the error as soon as I open the file (or reopen it)
[04:20:29] <ds3> the exact pop up is: Near line 19 of /tmp/gcode2r.txt: R clearance plane unspecified in cycle
[04:20:48] <cradek> can you pastebin the first few dozen lines of the file
[04:21:00] <ds3> sure
[04:22:57] <ds3> http://pastebin.ca/316966
[04:23:46] <cradek> ok
[04:23:55] <cradek> the key part of the spec is "several lines of code IN A ROW"
[04:24:07] <ds3> ahhhh
[04:24:48] <cradek> that's why it worked when I just used those two lines -- it breaks when I stick a G0 in between them
[04:25:44] <ds3> so then the debate is - is the current behavior sufficient or should it be modal
[04:25:56] <cradek> heh
[04:26:16] <cradek> not sure there's a debate - EMC implements RS274NGC (and this behavior seems right according to that)
[04:26:32] <SWPLinux> modal is a big problem waiting to happen
[04:26:36] <ds3> *nod*
[04:26:43] <SWPLinux> especially if it's modal across runs
[04:26:44] <ds3> SWPLinux: why is that?
[04:27:07] <cradek> if it's important to you to run these unmodified, I bet you can "fix" your EMC pretty easily
[04:27:14] <SWPLinux> because you may change the part, and then run a program that's poorly crafted (has no R spec), which will then use the old one
[04:27:22] <ds3> the way I was taught is - R plane is undefined. Either you specify it once or you have a disaster waiting to happen
[04:28:01] <SWPLinux> right. once perfile is good, once per set of holes is just as good (and how emc is now)
[04:28:08] <ds3> it isn't a big deal for me, just point out first impressions; R is modal for all the other ones I have tried
[04:28:31] <SWPLinux> modal things in emc are maintained (usually) across G-code files and program reset - anything not explicitly reset is left
[04:28:39] <SWPLinux> ... alone
[04:29:00] <SWPLinux> that's useful if you want to MDI something (like an R plane, for example), but not so good for safety in some cases
[04:29:11] <ds3> yes, that's why I was taught to have an 'insurance line' to clear out stuff
[04:29:31] <cradek> * cradek points vigorously at the ngc spec
[04:29:34] <cradek> haha
[04:29:44] <SWPLinux> always a good idea, IMO (though I've never programmed a CNC, so YMMV :) )
[04:30:15] <ds3> you just demostrated why CAM systems need so many different post processors even though they are all G code =)
[04:30:29] <cradek> yep, a royal pain
[04:30:49] <SWPLinux> heh
[04:30:55] <ds3> what I should do is put together a list of these findings, if anything just as a note to myself
[04:31:07] <cradek> does the haas support R-format arcs?
[04:31:21] <cradek> that seems like a big obstacle to making R be preserved forever
[04:31:50] <ds3> both R and I/J
[04:32:01] <ds3> R is good for hand programming ;)
[04:32:14] <cradek> so if you have a R format arc between the two G81, what happens to the canned cycle R plane?
[04:32:41] <ds3> you get a pile of scraps and some cheap tool holders appear on ebay? ;)
[04:32:46] <cradek> hahaha
[04:33:00] <ds3> I don't know. I can try it later in the term... still taking classes at the local JC and they have the HAAS machines
[04:33:06] <cradek> I thought maybe there are two different Rs
[04:33:15] <cradek> but it sounds a bit scary to me
[04:33:41] <ds3> just wait til the Lathe stuff gets implemented
[04:33:59] <cradek> in EMC? we have some already
[04:34:08] <cradek> and yeah, the threading canned cycle uses a lot of letters
[04:34:19] <tomp> jmkasunich: you around?
[04:34:21] <ds3> not the canned stock removal cycles
[04:34:23] <cradek> well, all of them, I think actually
[04:34:41] <cradek> right we may want to add one of those
[04:35:01] <SWPLinux> hey cradek - that reminds me. you were working on something to do roughing cycles on a lathe, right?
[04:35:06] <ds3> G71/G70 is a really neat thing for lathe code except every controller does it differently
[04:35:13] <cradek> my approach would be to make it the same as the threading cycle, but not synchronized
[04:35:26] <SWPLinux> at least, I remember seeing a screenshot of something that looked like that kind of toolpath
[04:35:41] <cradek> SWPLinux: that was the threading (synchronized) cycle
[04:35:58] <SWPLinux> hmmm - I think I remember something different
[04:35:58] <cradek> a stock removal cycle would look about the same
[04:36:00] <ds3> is should be easy enough to write a perl script to preprocess the G codes to expand the G70/G71
[04:36:04] <SWPLinux> kinda
[04:36:25] <cradek> ds3: in AXIS you can set up filters to preprocess different filetypes
[04:36:39] <cradek> the sim/axis setup has some examples
[04:36:39] <ds3> Oh?
[04:36:50] <ds3> okie
[04:37:02] <cradek> yeah you'd just make a filter for .haas that would monkey it around and spit out the NGC gcode to stdout
[04:37:17] <cradek> then you'll get auto translation when you load somefile.haas
[04:37:18] <fenn> i dont remember who it was that was looking at verot's sourceforge cvs, but you can get at the files if you roll back the date to before they were all deleted
[04:37:18] <ds3> hmmm that can solve my R plane stuff
[04:37:50] <cradek> if you make a "haas emulator filter", we'd happily put it in cvs
[04:38:13] <fenn> anonimasu/awallin
[04:38:44] <ds3> when I get to it. I has been at least a year since I wanted to mess with EMC. Just got some time recently
[04:39:07] <cradek> it has come a long way in the last year :-)
[04:39:51] <ds3> the only disappointment so far is it not working in a P100 w/16M :/
[04:40:01] <cradek> heh
[04:40:10] <cradek> forgive me for not crying too long about that
[04:40:31] <ds3> it all started with me wanting to update an old control with a PC104 board
[04:40:37] <cradek> I recently tried to run firefox on a 486 laptop...
[04:40:38] <tomp> awallin: hello
[04:41:11] <cradek> * cradek digs in his junkpile for a PIII motherboard for ds3
[04:41:43] <ds3> it needs to fit in the place of the original controller and looks like PC104+ is the closest formfactor
[04:42:04] <cradek> that won't be junkpile stuff then will it
[04:42:30] <ds3> I have PC104's in my junkpile but they are socket7 only... at best a 200MHz MMX
[04:42:32] <SWPLinux> I can suggest several PC104/PC104+ manufacturers if you like :)
[04:42:43] <ds3> can I get that in cash? ;)
[04:42:56] <cradek> SWPLinux: any boards we know work well with realtime?
[04:43:01] <SWPLinux> nope
[04:43:25] <SWPLinux> but Ray's mini-ITX experience may point the way for some others based on the same chipset
[04:46:20] <tomp> yesterday there was discussion about analog widgets for pyvcp. i began writing one and see i need to change direction. what was the downside to scaling the output ( like 1 unit of output = .001 or .1 or 10000 )?
[04:46:58] <cradek> think I'm off to bed, guys
[04:47:01] <cradek> goodnight
[04:47:06] <tomp> nite
[04:47:06] <SWPLinux> see you cradek
[04:47:50] <SWPLinux> tomp: was that the discussion about the jogwheel?
[04:47:56] <tomp> yes
[04:48:03] <tomp> any thoughts?
[04:48:21] <SWPLinux> soemone thought that it was imperative that a revolution of the wheel give an integer change in the output
[04:48:38] <SWPLinux> I dsiagree with that, and think a scale parameter/pin is a good thing
[04:48:50] <SWPLinux> but there's a snag, you see :)
[04:48:57] <tomp> ?
[04:49:18] <SWPLinux> this comes up with anything that can have the scale change with a nonzero output - you get a step function (possibly a big one)
[04:50:05] <tomp> big instantaneous changes? so will a numeric field entry
[04:50:13] <SWPLinux> true
[04:50:23] <SWPLinux> there's a way around it, I think
[04:50:31] <SWPLinux> hmmm - we could take this to -devel, since it's aprogramming issue
[04:50:39] <tomp> k
[04:53:02] <fenn> ds3: i think you are unlikely to find many pc104 systems that run faster than 200mhz
[04:53:26] <SWPLinux> not for less than $500 or so
[04:53:37] <fenn> not in x86 at least
[04:53:39] <SWPLinux> you can get P4 PC104+ cards though
[04:53:47] <SWPLinux> in fact, there's an Opteron card as well
[04:56:40] <fenn> pc104+ is ebx?
[04:57:12] <SWPLinux> no. EBX may be able to take PC104+ cards though (but I don't think so)
[05:01:50] <SWPLinux> hmmm. maybe that small Opteron card wasn't PC104
[05:02:06] <SWPLinux> or it was too damned expensive to keep making
[05:12:55] <ds3> fenn: the problem is enough memory; the ones i have can be coaxed into doing about 333MHz
[05:13:29] <ds3> I think at best I can squeeze in is 96M
[05:13:34] <fenn> you need more than that?
[05:13:44] <ds3> donno what EMC really needs
[05:13:53] <fenn> me either but i doubt it needs that much
[05:13:56] <ds3> and I not quite ready to run the experiment just yet
[05:14:07] <fenn> this was designed in like 1992 or so
[05:14:17] <ds3> the web pages says more like 400MHz and 192M or os is recommended
[05:14:20] <fenn> bah
[05:14:30] <fenn> that's just so ubuntu will be happy
[05:14:34] <ds3> Oh
[05:15:30] <fenn> i'm interested in making a lighter weight distro.. keystick or even no interface
[05:15:46] <ds3> that's what I want to try
[05:15:48] <fenn> send/receive nml over a lan connection to control the machine
[05:16:04] <fenn> so its more of a "black box" like a networked printer or something
[05:16:07] <ds3> there are apparently quite a few older machines that can use a better control (EMCO/ORAC to name a few)
[05:16:40] <SWPLinux> the Mesa 4C81+FPGA could be a good base for something like that, if RTAI can be coaxed to work on it
[05:16:55] <fenn> oh, you and your steppers..
[05:17:06] <SWPLinux> note the "+FPGA" in there ;)
[05:17:27] <ds3> it is more of me and my shallow wallet :/
[05:17:41] <SWPLinux> there's only one full-size IO connector there, plus a second with 12 I/Os (can be 6 differential pairs)
[05:17:43] <SWPLinux> heh
[05:18:00] <SWPLinux> the card is pretty reasonably priced, for an industrial buyer :)
[05:18:16] <SWPLinux> http://www.mesanet.com/cpucardinfo.html
[05:18:50] <ds3> ARM based? didn't know RT runs on ARM
[05:18:58] <SWPLinux> dunno if it does :)
[05:19:09] <fenn> i really doubt it does
[05:19:13] <ds3> hehe... I got a pile of old ARM boards if it does
[05:19:19] <SWPLinux> but the kernel has to be pretty stripped down for that, so RTAI may not be needed
[05:19:20] <fenn> that british guy was bitching about it a while ago
[05:19:28] <SWPLinux> it's not like there are 150 interrupt sources ...
[05:20:20] <SWPLinux> hmmm. there's supposed to be a Linux image for it, but I only see the NetBSD stuff on their site. bummer
[05:20:36] <ds3> fenn: any guesses as to what size of a hard drive foot print things can be squeezed down to?
[05:20:54] <fenn> the puppy rom was only like 40mb
[05:20:59] <fenn> but that assumes you have a crapload of ram
[05:21:25] <fenn> but it also had a lot of crap like web browsers and stuff
[05:21:34] <fenn> crap crap crap
[05:22:01] <fenn> i have 28 of these netier xl1000's and i'll be damned if i cant get it to run on them
[05:22:32] <SWPLinux> if you remove the X dependency by not needing any GUIs, and you don't need to compile on the unit, then I suspect it can get very small
[05:22:56] <ds3> sounds like a PC104+ board w/an AMD K6-2 333 + a CF card would make a nifty controller
[05:23:17] <fenn> yes that's essentially what i've got
[05:23:26] <SWPLinux> maybe. it depends on the RT performance of the CPU. not so much on throughput
[05:23:28] <fenn> it has normal pc ports though
[05:23:37] <fenn> parport vga usb etc
[05:24:22] <fenn> still havent tested rt latency
[05:28:30] <ds3> as long as it uses the same desktop chipset, why else would cause a difference in RT performance?
[05:28:56] <fenn> hardware stuff that can't be turned off
[05:29:07] <fenn> like.. wake on lan, uh.. i dunno what else
[05:29:18] <fenn> real time clock interrupt?
[05:29:53] <fenn> shared video memory is supposed to be bad
[05:31:26] <ds3> gotca
[05:31:36] <ds3> of course me and my steppers will tolerate that ;)
[05:32:08] <SWPLinux> steppers are the least tolerant of that stuff
[05:32:20] <SWPLinux> unless you're trying to run at exceedingly low step rates
[05:32:46] <fenn> * fenn hates steppers.. if you didnt already know
[05:33:17] <tomp> anyone got any links explaining hal components?
[05:33:43] <fenn> the source for blocks.c explains a lot of them
[05:33:49] <ds3> oh
[05:34:02] <tomp> thanks
[05:34:13] <fenn> i think i copied it into the wiki somehwere
[05:34:29] <fenn> but might have deleted that since two different places for the same thing is a bad idea
[05:34:50] <tomp> great, i tried reading the .comp files
[05:34:50] <ds3> make the wiki checkout the source ;)
[05:35:23] <jmkasunich> tomp: most of the newer ones (the ones that use comp) have manpages
[05:36:55] <tomp> will check thanks
[05:40:05] <tomp> found the man pages, but my system didnt... how do i connect those man pages so i can type 'man ddt' ?
[05:41:16] <SWPLinux> some form of man -P, I think
[05:41:28] <SWPLinux> try man man for fun :)
[05:41:59] <SWPLinux> oops - man -M
[05:42:07] <tomp> i was in man man now, it's mandb that i gotta grok
[05:42:22] <tomp> man man works
[05:42:38] <SWPLinux> man -M /Project/emc2/docs/man/ halcmd
[05:42:48] <SWPLinux> substituting your path, of course
[05:43:04] <SWPLinux> and the comp name
[05:47:03] <tomp> thanks, that works, but i gotta update the mandb... i'll never remember all that :)
[05:47:10] <SWPLinux> heh
[05:48:33] <SWPLinux> you could always source scripts/emc-environment, which will set the manpath for you
[05:48:38] <ds3> Why is RTAPI defined in a simulator build dependancy checking?
[05:49:19] <SWPLinux> RTAPI is still used in the simulator build, but the backend is a non-kernel thing
[05:49:35] <SWPLinux> the fact that it still includes proc.h is probably a bug thoough
[05:50:55] <ds3> oh. what confused me was the comment about ULAPI on top of hal_lib.c
[05:52:05] <SWPLinux> ah. that's because you can have either user or kernel modules with HAL (else you'd have a hard time writing halcmd, for instance)
[05:52:26] <SWPLinux> RTAPI also gives a minimal RT interface to RTAI or RTLinux (and simulator, of course)
[05:52:53] <ds3> gotcha. though it was a either or thing
[05:53:25] <ds3> thought
[05:54:19] <SWPLinux> it's more of an either / and thing :)
[05:58:07] <ds3> so many things to learn ;)
[06:07:09] <jmkasunich> bedtime
[06:07:15] <tomp> SWPLinux: does this sound right? if the jogwheel was reduced to just a 'ticker' then the analog pot is just wiring comps together
[06:07:26] <tomp> nite john, thanks
[06:07:29] <SWPLinux> see you jmkasunich
[06:08:02] <jmkasunich> tomp: there is still benefit to putting some or all of the analog part in the widget - saves HAL work
[06:08:51] <jmkasunich> take a look at the "knob2float" hal component, it does what you are doing (you are doing it for an on-screen jogwheel, knob2float does it for a real one using encoder counts)
[06:09:31] <jmkasunich> src/hal/components/knob2float.comp
[06:09:51] <jmkasunich> now I'm gone for real
[06:10:14] <tomp> bye,thanks, yes, surprised me, it's in the man9 dir
[06:11:04] <SWPLinux> hmm. I don't see the mechanism by which the output of the jogwheel is sent to the HAL pin
[06:11:14] <SWPLinux> of course, I don't understand python, so that's a detriment
[06:12:36] <SWPLinux> ah - self.count is the output value
[06:13:11] <tomp> yes, self.count this is all done already, it's just wiring
[06:13:38] <SWPLinux> is there a scale pin?
[06:13:46] <tomp> jmk has max & min ( so the pot doesnt go forever )
[06:13:53] <SWPLinux> or are you saying it's done because you've just about finished it? :)
[06:14:04] <tomp> scale yes ( on knop2float )
[06:14:20] <tomp> it finished me :)
[06:14:37] <SWPLinux> that component is meant for something that can change arbitrarily, but it's simpler for the software jogwheel
[06:14:38] <tomp> no i mean an example is needed, nothing to write
[06:14:58] <SWPLinux> I think it's best to add it to the widget
[06:15:26] <SWPLinux> by default, it wouldn't change anything about the jogwheel, but it can eliminate many lines in a hal file
[06:16:42] <tomp> can this new widget use the existing component? ( i didnt any comps in pyvcp_widgets.py )
[06:17:11] <SWPLinux> no. that component is a kernel-space realtime component
[06:17:21] <SWPLinux> pyvcp is very much userspace
[06:19:00] <tomp> pyvcp_widgets is where the widget classes are
[06:19:26] <SWPLinux> yep. I'm editing there right now - trying my luck with the snake :)
[06:19:46] <SWPLinux> I'm having a harder time creating the HAL pin than the changes to the widget are :)
[06:22:20] <tomp> adding the scale to the jogwheel? ( default to unity for digital use?) adding min&max ?
[06:22:39] <SWPLinux> scale only
[06:23:01] <SWPLinux> for completeness, there should probably be a reset pin
[06:23:12] <SWPLinux> min and max could be good, and are also trivial to add
[06:23:16] <tomp> infinite turn pot :)
[06:23:19] <SWPLinux> heh
[06:23:42] <SWPLinux> actually, a set and a reset pin would be nice - so you could set the value to the current axis position, for example
[06:25:19] <tomp> maybe arm (like enable): turn pot to value THEN connect output to system ( no mouse bumped spindle speed )
[06:25:46] <tomp> oh preset, yes
[06:25:59] <SWPLinux> yeah. you pretty much need all the stuff jmk stuck in his, but it probably shouldn't need all those HAL connections to use
[06:26:37] <tomp> maybe some of his default... i oughta read more
[06:27:37] <tomp> but there's soft-pullups :) a const connected to an unneeded input
[06:28:25] <SWPLinux> yeah :)
[06:31:29] <tomp> wow, the 'enable' is red paint on the brass pot screw ( dont touch this pot or your warranty is void )
[06:32:21] <SWPLinux> if enable is off, it should still track movement, but shouldn't change the output (basically it setst the scale to 0)
[06:34:18] <tomp> yes, if the implementation was by wiring components, the widget display would change, but the end effect would be fixed (bad feedback)
[06:34:51] <tomp> the widget should know if it's being governed
[06:35:07] <SWPLinux> no, I'm saying that the widget can have the dot movea round on the knob, but its output shouldn't change at all
[06:35:19] <SWPLinux> err - yes, maybe we're saying the same thign
[06:35:21] <SWPLinux> thing, even
[06:35:49] <tomp> right, the user should know if he's being ignored
[06:36:18] <tomp> did you change the jogwheel?
[06:36:42] <SWPLinux> I've been doing some editing, but I've found that I can't run pyvcp at all, so I'm re-checking out right now
[06:39:22] <SWPLinux> hmmm. it looks like -z9 doesn't help much in checkout speed. maybe -z5 would have been better
[06:41:08] <A-L-P-H-A> SWPLinux, got time for a philisophical discussion (not even a debate)
[06:41:29] <tomp> :) i got the idea of wiring it all up in hal from ruggadallur's (?) torchheight example.... i think your take on that is... that's a mess of a diagram to wire up, make simpler unified thing to do it
[06:41:31] <SWPLinux> some, though I'm getting tired, so I may not be of much help/interest :)
[06:41:59] <SWPLinux> tomp: in cases where adding features doesn't specialize the component, I'm pretty much for it
[06:42:10] <SWPLinux> yay - the checkout is done
[06:42:48] <A-L-P-H-A> okay... lots of the world hates America... what it does, it's stereotypical citizen, and mainly it's government. What changes would America need to do, so that it's no longer seen as a social outcast that it makes itself out to be.
[06:42:49] <tomp> gotcha... generalize it
[06:43:12] <SWPLinux> yay - the compile is done :)
[06:43:23] <A-L-P-H-A> Some lump anything American into that pile... but there are nice people, like you, and others included in this chan...
[06:43:47] <SWPLinux> A-L-P-H-A: I think we'd basically have to act like a responsible world citizen for a decade or two
[06:43:48] <A-L-P-H-A> How can some people in your country elect a monkey into office? I mean, his entire time there, he's taken the blame for something once.
[06:44:04] <SWPLinux> he did?
[06:44:10] <SWPLinux> I don't remember that
[06:44:14] <A-L-P-H-A> He DID! recently... I was so fuck'n surprised.
[06:44:32] <SWPLinux> wow. I've been avoiding the news for the last few years because it's so depressing
[06:44:52] <SWPLinux> ok. regarding the election:
[06:44:57] <SWPLinux> elections, actually
[06:44:59] <A-L-P-H-A> http://www.msnbc.msn.com/id/9324891/ he took the blame for a few things...
[06:45:17] <SWPLinux> the problem is that in the US, a lot of people don't want to have to think too much
[06:45:21] <A-L-P-H-A> also, recently, the big one, that he took the blame for was the # of soldiers in iraq was too low, and it was his fault.
[06:45:37] <A-L-P-H-A> seriously, I rather be living in German, that in the US.
[06:45:43] <A-L-P-H-A> but so glad I'm in socialist Canada.
[06:45:57] <SWPLinux> there's a pretty strong anti-knowledge, anti-learning thing going on here
[06:46:14] <A-L-P-H-A> IL, was taking creationism as fact.
[06:46:17] <A-L-P-H-A> that's pretty sad.
[06:46:33] <A-L-P-H-A> every country has it's problems...
[06:46:39] <SWPLinux> I assume you've seen the bumper stickers that say something like "my child is an honor student at <school>"
[06:46:54] <A-L-P-H-A> but most people in the world, if they know Canada, the like us... Same with say Sweden, or Finland.
[06:47:03] <SWPLinux> well, we also have ones here that say "my kid beat up your honor student"
[06:47:19] <A-L-P-H-A> But mention, North Korea, US, any middle eastern country, there's disdain.
[06:47:44] <SWPLinux> sure - as a policy, the US tends to promote its own agenda, and to sabotage that of other countries
[06:47:47] <SWPLinux> it's hard to respect that
[06:48:05] <A-L-P-H-A> very much so.
[06:48:06] <SWPLinux> and the agenda has to do with big business and "personal" things
[06:48:23] <SWPLinux> personal as in "you're my friend and I have power, so we'll do stuff to help each other out"
[06:48:34] <A-L-P-H-A> prop up a leader you can control... once outta control, invade their country, and set up a kangaroo court to hang him... oh and release a video of it on youtube.
[06:48:59] <SWPLinux> well, I wouldn't blame all of that on us
[06:49:14] <SWPLinux> you've got to remember - we just happened to outlast the Soviet Union
[06:49:47] <SWPLinux> almost everything that happened in US and USSR foreign policy from around 1946 on was part of the cold war
[06:50:15] <SWPLinux> the space program was something we did so we could beat the Soviet Union to the moon ...
[06:50:44] <A-L-P-H-A> seems like you were started to be hated in the mid 80s but random countries as well.
[06:50:50] <A-L-P-H-A> early 80's.
[06:51:03] <A-L-P-H-A> Iran inncident, where the Canucks bailed your hostages out.
[06:51:03] <SWPLinux> the Korean War, Vietnam War, and lots of stuff in the middle east have been the US and USSR fighting proxy wars
[06:51:10] <A-L-P-H-A> true.
[06:51:29] <A-L-P-H-A> I would, as a rational person, blame America... 1.) the american public someone appointed a non-elected president in his first term. Elected him on the second term. 2.) have him not impeached by the senate or house [however your system works]. 3.) allow Bush to do so much shit that it just isn't funny.
[06:51:34] <SWPLinux> now that there's no USSR, there's a lot of shit to clean up, and it's all our fault ...
[06:52:05] <SWPLinux> it's not all our fault, but we're the only ones left to blame for it (and it's partly our fault, for sure)
[06:52:12] <A-L-P-H-A> Mind you the choices you had to elect was a turd or a douchebag.
[06:52:18] <A-L-P-H-A> turd won.
[06:52:23] <SWPLinux> it makes no difference
[06:52:30] <A-L-P-H-A> yeah...
[06:52:36] <SWPLinux> does your election system work like that of the UK?
[06:52:52] <A-L-P-H-A> Dean would have made an interested crazy president... I sure think he wished he didn't make that war cry.
[06:53:05] <A-L-P-H-A> SWPLinux, yes... where we can fire the primeminister.
[06:53:10] <A-L-P-H-A> Parliment...
[06:53:10] <SWPLinux> he didn't - that was a sensationalized bit of media fuckup, if you ask me
[06:53:17] <A-L-P-H-A> yup.
[06:53:17] <SWPLinux> Parliament
[06:53:30] <SWPLinux> he's a nice guy, and did a lot of good here
[06:54:00] <A-L-P-H-A> If Obama runs, I hope he wins... I think he said he's running... if not yet... just not the words yet.
[06:54:07] <SWPLinux> he was a practicing doctor here before he became Governor
[06:54:12] <SWPLinux> (Dean, that is)
[06:54:15] <A-L-P-H-A> Ahh.
[06:54:23] <A-L-P-H-A> Senator Clinton... argh... :(
[06:54:41] <A-L-P-H-A> Seriously, who's up to take over Bush's job?
[06:54:49] <A-L-P-H-A> Rep side.
[06:55:03] <SWPLinux> in this state, he managed to: balance the budget - we actually had a surplus for a few years; improve healthcare for lots of people; improve educational grants and that kind of thing ...
[06:55:14] <SWPLinux> I have no idea at this point
[06:55:25] <SWPLinux> I think Obama won't get elected because he's not white
[06:55:31] <A-L-P-H-A> Obama is smart, charismatic, and has the support of the people... so unless he's a closet fag, or molested little chidlren, kinda hard right now to see him not win... unless the country is still racist in so many rights.
[06:55:39] <SWPLinux> there's a lot of racist sentiment in the US, sadly
[06:55:55] <A-L-P-H-A> yup
[06:56:17] <A-L-P-H-A> Canada has had a female Prime Minister... albeit non-elect. Appointed.
[06:56:29] <A-L-P-H-A> UK with Margaret Thatcher...
[06:56:32] <SWPLinux> he seems like an interesting guy. He gave the speech when my wife got her Masters degree last June
[06:56:43] <A-L-P-H-A> Cool...
[06:57:21] <A-L-P-H-A> seriously, I think a lot of the world would respect the US more, if they had a guy like him in power.
[06:57:36] <SWPLinux> as far as teh reputation of the US, I think there's no quick answer. building trust takes a long time, but destroying it takes only an instant
[06:58:02] <A-L-P-H-A> Takes one or two presidents.
[06:58:13] <SWPLinux> takes one day with one president
[06:58:19] <SWPLinux> to kill it
[06:58:19] <A-L-P-H-A> but I don't think the US was hated that much with Bill Clinton in power...
[06:58:23] <SWPLinux> several terms to get it back
[06:58:39] <SWPLinux> no - Clinton actually had a pretty responsible foreign policy
[06:58:43] <A-L-P-H-A> but I was also younger then, and didn't follow politics as much.
[06:58:54] <SWPLinux> Bushj didn't know the words "foreign policy"
[06:59:01] <A-L-P-H-A> He seemed to keep the status quo... a good thing.
[06:59:08] <K`zan> Heh, obama got nothing to be elected for. Newbie who has done nothing to prep him for pres.
[06:59:14] <SWPLinux> Clinton?
[06:59:23] <K`zan> ewwwwwwwwwww
[06:59:25] <A-L-P-H-A> K`zan... go back to your tree fort.
[06:59:34] <K`zan> Here "=_
[06:59:35] <A-L-P-H-A> or under the bridge. :)
[06:59:41] <K`zan> There too
[06:59:44] <SWPLinux> no - I was asking if it was Bill Clinton that A-L-P-H-A weas saying maintained the status quo
[06:59:55] <A-L-P-H-A> SWPLinux, yes.
[07:00:00] <SWPLinux> I'm not sure Hillary is any good for pres either
[07:00:02] <K`zan> Blowjobs are not status quo?
[07:00:09] <SWPLinux> they are for powerful people
[07:00:14] <SWPLinux> or those with Jaguars ;)
[07:00:18] <A-L-P-H-A> K`zan, that's his private life... I honestly couldn't care about his private life.
[07:00:37] <K`zan> I don't do my private life at the office A-L-P-H-A - BIG difference
[07:00:51] <A-L-P-H-A> K`zan, he lives in the same fuck'n building.
[07:00:53] <A-L-P-H-A> that's his home.
[07:00:56] <K`zan> Irrelivant.
[07:01:10] <SWPLinux> he screwed up (no pun intended), and made it even worse by lying about it
[07:01:15] <K`zan> No, he lives in his office.
[07:01:24] <K`zan> Yep, man has no integrity at all.
[07:01:29] <SWPLinux> that is absolutely insignificant comapred to the colossal cluster-fuck that is the Bush presidency
[07:01:30] <A-L-P-H-A> K`zan, your morales can be in alignment with the other small minded americants. :)
[07:01:48] <A-L-P-H-A> hey alex_joni.
[07:01:52] <K`zan> Bush is not jewel, but a LONG way from a clusterfuck
[07:01:54] <SWPLinux> hi alex_joni
[07:02:10] <K`zan> A-L-P-H-A: you can be so pathetic sometims, sigh...
[07:02:14] <SWPLinux> well, if Bush isn't a cluster-fuck, I'm not sure what it would take for you to see one
[07:02:17] <A-L-P-H-A> K`zan... a sad republicant you are. :)
[07:02:37] <K`zan> Clinton, carter... Good examples of clusterfucks.
[07:02:52] <A-L-P-H-A> This is a democrat, socialist , liberal area... :)
[07:02:59] <SWPLinux> well, I gues sit depends on your definition of good
[07:03:06] <K`zan> A-L-P-H-A: No, not so, I have just done more than live in moms basement in fantasyland :).
[07:03:07] <A-L-P-H-A> SWPLinux. :) heh
[07:03:37] <K`zan> I am personally of the opinion that a number of people have something serious on Bush.
[07:03:50] <SWPLinux> if making your friends rich, killing loads of innocent people, spltting the opinions of the country and the world, and bringing terrorism out of the closet is good, then Bush is a great president
[07:04:18] <K`zan> SWPLinux: Come on, terrorism was out of closet LONG before bush got here.
[07:04:22] <A-L-P-H-A> SWPLinux, aside from those. he's awesome! still acts like a monkey....
[07:04:26] <SWPLinux> oh, and increasing the national debt by $trillions per year
[07:04:46] <A-L-P-H-A> I wish I was Bush's best friend... I could be so much richer right now.
[07:04:50] <SWPLinux> I'm pretty sure that terrorism was a much lesser threat 10 years ago than it is now
[07:04:53] <A-L-P-H-A> but at the cost of my soul.
[07:04:57] <K`zan> SWPLinux: Check you facts, the debit is down. Get real info, not from the commercial liberal media :)
[07:05:22] <SWPLinux> I know there were hijackings every other day in the 70's, but it seemed to settle down a lot between then and the 2000's
[07:05:23] <K`zan> It really amazes me how few people actually check out what they are told.
[07:05:55] <K`zan> SWPLinux: Ever hear of the arabic term "Hudna"?
[07:06:00] <A-L-P-H-A> K`zan, your wrong. yet again.
[07:06:00] <A-L-P-H-A> http://en.wikipedia.org/wiki/U.S._public_debt
[07:06:20] <K`zan> wikipedia is KNOWN to be hosed.
[07:06:36] <K`zan> Any idiot can change it and many do :)
[07:06:37] <SWPLinux> og course, you won't believe this: http://www.brillig.com/debt_clock/
[07:07:07] <SWPLinux> or this: http://www.cedarcomm.com/~stevelm1/usdebt.htm
[07:07:17] <A-L-P-H-A> http://www.publicdebt.treas.gov/opd/opds012006.htm GOV site.
[07:07:19] <A-L-P-H-A> it's factual.
[07:07:35] <K`zan> I cna write a script that adds on a daily basis too :).
[07:07:41] <SWPLinux> or this: http://www.federalbudget.com/
[07:07:55] <K`zan> It only recently started decreasing from what I have read.
[07:07:57] <A-L-P-H-A> I just gave you the TREASURY DEPT!!!! For fuck sakes, that's correct.
[07:08:12] <K`zan> .com? I don't think so...
[07:08:15] <A-L-P-H-A> NO!
[07:08:17] <SWPLinux> you have to question those news sources, don't cha know
[07:08:18] <K`zan> .gov yes
[07:08:27] <SWPLinux> http://www.publicdebt.treas.gov/opd/opdpenny.htm
[07:08:35] <K`zan> Yep, I spend an inordinate amount of time doing that...
[07:08:45] <K`zan> Too much probably :-)
[07:08:48] <A-L-P-H-A> my God, K`zan... you're a little thick today.
[07:09:08] <K`zan> A-L-P-H-A: No Sir, you are a little brainwashed and stuck on stupid :)
[07:09:25] <A-L-P-H-A> the numbers don't contradict wikipedia... so how is that wrong?
[07:09:36] <K`zan> Sigh, usless talking to closed brainwashed minds, sigh...
[07:09:51] <SWPLinux> neither of you is particularly stupid or brainwashed. you just disagree and are unwilling to accept that the other has a valid point which could be backed up by factual data
[07:10:17] <K`zan> Too many points of "factual data" sometimes.
[07:10:29] <SWPLinux> I'm in the same boat. I assume that K`zan gets her information from Fox news and the various conservative blogs/news sites
[07:10:47] <SWPLinux> whereas it's assumed that I get my information from "the liberal media"
[07:10:53] <K`zan> No only, but the loons are pretty easy to spot on the other side too.
[07:11:04] <K`zan> Not
[07:11:25] <SWPLinux> that's the problem - when someone says "bush is an idiot", the response is "well, clinton was a slut"
[07:11:29] <A-L-P-H-A> K`zan, you're not by chance... ooooooooooooh. shit... who's that female version of Rush Limbagh (sp)....
[07:11:39] <SWPLinux> kinda changes the subject
[07:11:43] <A-L-P-H-A> Jymmm... republican or democrat?
[07:12:11] <SWPLinux> heh - he's brilliantly letting that one go :)
[07:12:22] <K`zan> When it comes right down to it, there is too much smoke and mirrors and agenda and commercial driven science to be real sure of anything you hear anymore :-(.
[07:12:26] <A-L-P-H-A> Ann Coutcher... I think.
[07:12:31] <SWPLinux> jsut for the record, my answer is "neither"
[07:12:43] <A-L-P-H-A> Ann Coulter
[07:13:01] <A-L-P-H-A> Sensationalist republican.... Ann Coulter.
[07:13:09] <SWPLinux> I actually voted for democrats, republicans, green party, and independents in the last election (maybe others as well - I don't remember)
[07:13:13] <K`zan> I'm just an American, as far as I am concerned both parties are so far past fucked in the head it is pathetic.
[07:13:19] <SWPLinux> hmmm - I think there was a libertarian in there
[07:13:41] <K`zan> We need to have 20 parties like Canada, keeps them all confused :-).
[07:14:00] <K`zan> Seriously, two parties are essentially the same.
[07:14:12] <A-L-P-H-A> I'm pretty sure there's like 4... maybe 5. I could be wrong.
[07:14:29] <SWPLinux> it's funny. on an internet forum (several years ago), someone who tends to be conservative said soemthing like "MoveOn.org was founded by a wealthy democrat", like it was fact
[07:14:40] <K`zan> A-L-P-H-A: Somewhere in there and I think that is a far better thing than 2.
[07:14:56] <A-L-P-H-A> K`zan... most european countries have multiple, don't they?
[07:15:24] <SWPLinux> I responded that no, it was actually founded by several young people who wanted to just get the clinton impeachment over with quickly ("censure and move on" was the message, hence the name)
[07:15:27] <K`zan> Don't think as many as .ca though, not sure to be honest - got enough problems here to keep me occupied :).
[07:15:43] <A-L-P-H-A> Ross Perot... did he fuck up the electrions for GB Sr? Or was there no hope in hell for GB Sr anyways?
[07:16:01] <SWPLinux> I also asked him for the source of his information, and mentioned that I have met (and my wife knows through work) the mother of one of the organizers, so my information is pretty good
[07:16:12] <SWPLinux> he never responded to me
[07:16:40] <SWPLinux> that happens whenever he tangles with people who really know the subjects he likes to spout about
[07:16:41] <K`zan> Until we are willing to apply the "rope, politician, lamp post - some assembly required" we are not going to get better, sad to say.
[07:16:53] <A-L-P-H-A> huh? So my cousin's best friends' dog sitter's bf said so?
[07:16:59] <SWPLinux> something like that
[07:17:13] <SWPLinux> some disassembly required ;)
[07:17:44] <SWPLinux> I wish it were possible to have "a vote of no confidence" here
[07:17:51] <K`zan> As my grandmother once told me: "Don't believe anything you hear and only half of what you see". Not sure about the last part anymore with the ability to manipulate media that we have now.
[07:17:53] <SWPLinux> but we don't have that
[07:18:11] <SWPLinux> well, see for real and see in the news are two different things
[07:18:15] <K`zan> Gads, we are supposed to have more snow...
[07:18:23] <A-L-P-H-A> K`zan... how does someone's sex life affect how the country is ran? Honest question... not trying to bait you.
[07:18:46] <K`zan> Problem with the whole political arena is that there are no penalties for fucking up.
[07:19:20] <SWPLinux> A-L-P-H-A: I can answer that one (with the stupid political shit left out): the president is supposed to represent an nhonest upstanding citizen - to be an example for everyone in the US and ideally elsewhere in the world
[07:19:21] <K`zan> A-L-P-H-A: In ANY business you got caught doing what klintoon pulled, you would be out on your ass and unemployable...
[07:20:00] <K`zan> Coffee, brb.
[07:20:01] <SWPLinux> when they betray a trust (which marriage is, as is the presidency) they should be held accountable
[07:20:03] <A-L-P-H-A> K`zan... only if you get caught.
[07:20:08] <SWPLinux> wrong!
[07:20:15] <K`zan> A-L-P-H-A: That is SO wrong...
[07:20:20] <SWPLinux> laws are there whether you get caught or not
[07:20:40] <A-L-P-H-A> SWPLinux... I'm only guilty if convicted of said offense against the law.
[07:20:43] <SWPLinux> if you won't follow laws you don't like, why would you expect anyone else to?
[07:20:47] <SWPLinux> bullshit
[07:20:50] <K`zan> As a cop once told me: everyone is guilty, we just haven't caught you yet :-=)
[07:21:19] <SWPLinux> you're only punishable by law if you get caught and convicted, but you're guilty anyway
[07:21:30] <A-L-P-H-A> SWPLinux. no.
[07:21:48] <K`zan> And if you have ever seen one of the major law libraries, you won't doubt it :-)
[07:21:49] <A-L-P-H-A> guilty is only when convicted... you're innocent until proven guilty.
[07:22:05] <SWPLinux> yeah yeah - tell it to the judge
[07:22:07] <K`zan> Bullshit
[07:22:19] <A-L-P-H-A> SWPLinux... peers. Jury.
[07:22:19] <SWPLinux> if you can't self-regulate, and 99% of the world is like you, then we're all fucked
[07:22:31] <K`zan> Lawyers manipulate the law daily and the guilty walk..
[07:22:48] <A-L-P-H-A> K`zan... guilty and being really responsible for the crime are different.
[07:22:56] <SWPLinux> money talks and bullshit walks, as the saying goes
[07:22:59] <A-L-P-H-A> you're using the word incorrectly, that's all.
[07:23:15] <SWPLinux> wrong. there are multiple meanings to many english words
[07:23:29] <SWPLinux> that's why I said "legally punishable" back there
[07:23:39] <K`zan> Quick little story?
[07:23:48] <K`zan> Wanna hear it?
[07:23:49] <SWPLinux> go fer it
[07:24:08] <A-L-P-H-A> I'll read it when I come back... I'm gonna reheat some pizza. :)
[07:24:13] <K`zan> Years ago now I was on a job down in Miami (traveling on business).
[07:24:46] <K`zan> Job ran long and rather than fly home to Atlanta and back on Monday, I decided to spend the weekend with friends there...
[07:25:37] <K`zan> Sunday morning we were heading out to a nice place for a (expense account :-) brunch...
[07:26:36] <K`zan> On the way back down the driveway to the apartments (longish and windey) there is a spanish woman walking down the driveway...
[07:27:09] <K`zan> about 50 yards or so in front of us...
[07:27:51] <K`zan> Black guy jumps out of some bushes and knocks the woman down and beats the shit out of her and steals her purse...
[07:28:12] <K`zan> I take off in the van to get there to help...
[07:28:25] <K`zan> Black guy takes off across the lawn (scrub really)...
[07:28:59] <K`zan> Pat jumps out to see if the woman was OK (the black guy REALLY beat the hell out of her)....
[07:29:08] <K`zan> I take off after him in the van...
[07:29:31] <K`zan> At that point I know who he is and sure as hell know he is guilty...
[07:29:56] <K`zan> Long story short, he dodges between two buildings the van won't fit through...
[07:30:06] <K`zan> I find a way around...
[07:30:30] <K`zan> Come out and spot the guy sitting in a carport (shiney black from the exercise)...
[07:31:11] <K`zan> At that point I can not prove in a court of law that he is the one, his fellow racemates tell me he has been sitting there for the last hours....
[07:31:18] <K`zan> hour
[07:31:42] <K`zan> So, I guess at that point, legally, he is not guilty...
[07:32:33] <K`zan> 4 of them (black, FWIW) and two of us back when looking at a black was racism / discrimination / whatever...
[07:33:04] <SWPLinux> right. legally, he's presumed iunnocent until you can convince a jury (ideally a non-biased one) that he's the one who did it
[07:33:17] <K`zan> Guilt does not necessarily have anything to do with the law...
[07:33:38] <SWPLinux> that's the main word missing in A-L-P-H-A's sentence. it's "presumed innocent until proven guilty", not "actually innocent until proven guilty"
[07:34:00] <SWPLinux> and afterwards, it's "acquitted"
[07:34:02] <K`zan> Yep, and too many people are unable to understand that difference.
[07:34:06] <SWPLinux> (or better, convicted)
[07:34:10] <K`zan> Heh
[07:34:41] <K`zan> Watch the "legal system" here and you will indeed loose heart and confidence.
[07:34:53] <SWPLinux> too much politics in the legal system as well, even at a local level
[07:35:03] <SWPLinux> I kinda liked the one time I went to court
[07:35:04] <K`zan> Yep, dunno what the answer is..
[07:35:21] <SWPLinux> story, if you like :)
[07:35:25] <K`zan> The old west might have been as peaceful as history doesn't portay it.
[07:35:29] <K`zan> Go for it...
[07:35:48] <SWPLinux> well, my mother had been in an auto accident
[07:36:03] <SWPLinux> naturally she called me, since I'm the one she always calls ...
[07:36:40] <SWPLinux> this was at an intersection in town, and the other car had skimmed past her
[07:37:22] <SWPLinux> there were 13-foot skid marks from where she had hit the brakes, and her car had been pushed a couple of feet to the right. her airbag also went off, injuring her and the car
[07:37:40] <SWPLinux> the other person hadn't applied the brakes at all, and happened to be traveling uphill
[07:38:44] <SWPLinux> anyway, a year or so goes by, and my mother finds out that she's being sued by the other lady's insurance company (as it turned out, my mother's insurance had lapsed because a bill had gotten lost in the mail, which she rectified as soon as she was told by the police)
[07:38:46] <K`zan> Let me guess, judge nailed your mother.
[07:38:56] <SWPLinux> the other insurance company was trying to get her on that technicality
[07:39:02] <SWPLinux> bzzzzt! wrong :)
[07:39:18] <K`zan> Yep, heard way too many stories like that. NOT GOOD>
[07:39:45] <SWPLinux> actually, it was funny. I suggested that my mother prepare a bill for ther damages her car had sustained, and also to write up her memory of the accident
[07:39:56] <K`zan> The legal system is too corrupt and what to do about it I just do not have an answer for, other than a violent and nasty one.
[07:40:29] <SWPLinux> anyway, when the pricey lawyer had finished, and my mother had just calmly told the judge what had happened, the judge basically turned to the other lady and said:
[07:40:52] <SWPLinux> "here's what happened. you saw a green light, but it was red by the time you got to it. pay the lady"
[07:40:59] <K`zan> Sorry, thought you were done...
[07:41:22] <K`zan> Sometimes it works out, glad your mom got a break.
[07:41:22] <SWPLinux> so my mother ended up getting the money for her damages, and the other lady, the lawyer, and the insuracne company had to go suck eggs :)
[07:41:34] <SWPLinux> yeah - that was kind of the opposite story - sometimes it does work
[07:41:50] <K`zan> I have one more, if you want to hear it...
[07:41:58] <SWPLinux> soap - ballot - jury - ammo ,,,
[07:42:09] <K`zan> :-)
[07:42:12] <SWPLinux> sure
[07:42:18] <K`zan> Back in Atlanta...
[07:42:32] <K`zan> Going around 285 (I forget why now)...
[07:42:45] <K`zan> Parking lot ahead, slow down, stop....
[07:42:56] <K`zan> Car plows into the back of me...
[07:43:17] <K`zan> Not real hard, but def noticible - no damage to either vehicle.
[07:44:06] <K`zan> Black woman in the car calls cops. Ok, well whatever...
[07:44:16] <K`zan> Cops FINALLY get there....
[07:45:06] <K`zan> The cop (husky black dude :-) and I are standing between cars looking at.... nothing...
[07:46:01] <A-L-P-H-A> okay.
[07:46:02] <A-L-P-H-A> back.
[07:46:07] <A-L-P-H-A> read most of it...
[07:46:12] <Jymmm> SWPLinux array shoft
[07:46:15] <Jymmm> shift
[07:46:26] <K`zan> Woman jumps out of car and starts screaming (literally) to the cop - that racist bitch locked up the brakes and made me hit her, she hates blacks and is discriminating against me (this goes on for a ludicirisly long time)....
[07:46:50] <K`zan> She winds down and looks at the cop...
[07:46:57] <K`zan> I figure I am had...
[07:47:25] <K`zan> Cops looks at her and explains that she hit me and writes her a ticket and tells her to knock off the racist bullshit...
[07:47:33] <K`zan> He tells me I can leave :-).
[07:47:45] <SWPLinux> heh - nice :)
[07:47:56] <SWPLinux> Jymmm: what?
[07:47:57] <Jymmm> K`zan I would thrown your ass in jail
[07:48:09] <SWPLinux> jymmm - you're a white-hater and you know it
[07:48:11] <K`zan> Warmed the heart in Atlanta :-).
[07:48:28] <Jymmm> SWPLinux Yep, that be me!
[07:48:37] <K`zan> That is where I expected to go if I was lucky. Got a real human being for a cop - rare.
[07:48:42] <SWPLinux> shee-it. what'choo wanna shift a array for?
[07:48:50] <Jymmm> SWPLinux: soap - ballot - jury - ammo
[07:48:53] <K`zan> Anyway, there is good and bad...
[07:48:57] <SWPLinux> it's pretty common here
[07:49:05] <A-L-P-H-A> okay... fun stories.
[07:49:19] <K`zan> I'm freezing tonight....
[07:49:21] <A-L-P-H-A> wish I had more pizza.
[07:49:33] <Jymmm> K`zan Nah there are some decent cops out here. They just dont like drama.
[07:49:33] <A-L-P-H-A> K`zan... you've burnt out your soul?
[07:49:37] <K`zan> Me too :-), got munches.
[07:49:39] <SWPLinux> I've had cops pull up to me at a traffic light, motion for me to roll down my window, and say "hey - you've got a taillight out - you oughtta get that fixed"
[07:49:49] <A-L-P-H-A> Jymmm,not on my fuck'n street.
[07:50:06] <K`zan> SWPLinux: you are lucky, that is generally considered probable cause anymore...
[07:50:10] <SWPLinux> "thanks, officer" ;)
[07:50:33] <SWPLinux> well, around here, there's not a tlot of that
[07:50:35] <A-L-P-H-A> what a bitch ass cop... tried to scare me on my own street... out of his jurisdiction, and trying to intimadate me. I should have called police HQ, and reported his ass... as I used to work for Toronto Police...
[07:50:44] <SWPLinux> it's getting worse, like everywhere, but it's not bad yet
[07:50:43] <A-L-P-H-A> fuck'n asshole cop... that's on my street.
[07:50:51] <K`zan> No matter how much I don't like or trust the cops, I ***ALWAYS*** make sure I pass the attitude check :).
[07:51:13] <K`zan> Cops here are worse than the DeKalb county cops in atlanta.
[07:51:16] <A-L-P-H-A> Nah... I just name drop from time to time... :) Friends are nice.
[07:51:36] <A-L-P-H-A> UCPD... lets go get tasered!
[07:51:39] <SWPLinux> indeed. I've found that in basically every situation (the military being a notable exception), you get a lot more done by being nice than by being an asshole
[07:51:50] <K`zan> "We can't arrest anyone, it would take us off the streets"...
[07:51:57] <SWPLinux> and it even works in the army sometimes
[07:52:08] <K`zan> Was told that not once but 3 separate times.
[07:52:27] <SWPLinux> are you still in the Seattle area?
[07:52:32] <A-L-P-H-A> SWPLinux, you were in the army right?
[07:52:33] <SWPLinux> or atleast WA
[07:52:37] <alex_joni> morning guys
[07:52:37] <K`zan> You can always give them the finger under the dash after he lets you go.
[07:52:40] <A-L-P-H-A> hey alex_joni.
[07:52:41] <SWPLinux> I was in the national guard
[07:52:45] <K`zan> SWPLinux: Yes, scabattle...
[07:52:47] <SWPLinux> army national guard
[07:52:53] <A-L-P-H-A> SWPLinux, oh... could you be recalled to active duty at any time now?
[07:52:58] <Jymmm> K`zan lol you funny lady =)
[07:52:59] <SWPLinux> I suppose so
[07:53:07] <K`zan> SWPLinux: thanks for your service, appreciate it more than you know.
[07:53:13] <SWPLinux> they haven't called me yet
[07:53:17] <SWPLinux> you're welcome :)
[07:53:19] <A-L-P-H-A> SWPLinux. :( but aren't you over a prime age to be recalled... so unless they were desparate?
[07:53:34] <K`zan> Jymmm: I try to keep a sense of humor...
[07:53:35] <SWPLinux> no. that's another funy story (much shorter though)
[07:53:36] <A-L-P-H-A> SWPLinux... when's my check clear?
[07:53:37] <Jymmm> SWPLinux If they call you to active duty, you'ld have to hit the barber wouldn't you?
[07:53:42] <alex_joni> they are desperate if they recall him
[07:53:45] <alex_joni> regardless of age
[07:53:46] <alex_joni> :)
[07:53:52] <SWPLinux> dunno, but I'll be sending the boards out on Tuesday (Monday is a holiday)
[07:54:05] <A-L-P-H-A> SWPLinux, you haven't sent any out?
[07:54:11] <K`zan> If they would take me I'd sure go...
[07:54:11] <A-L-P-H-A> or just my set?
[07:54:12] <SWPLinux> just for that, Alex's boards aren't going out until Friday ;)
[07:54:17] <alex_joni> :P
[07:54:23] <alex_joni> you know that doesn't hurt me at all
[07:54:24] <alex_joni> :D
[07:54:24] <K`zan> I am a great sniper and love the work :).
[07:54:25] <SWPLinux> just the ones I ahdn't received checks for :)
[07:54:26] <Jymmm> SWPLinux Wait till Friday the 32nd.
[07:54:38] <A-L-P-H-A> SWPLinux, say again?
[07:54:40] <SWPLinux> the 6th tuesday of the month
[07:55:10] <K`zan> SWPLinux: you do boards?
[07:55:12] <SWPLinux> I sent out a bunch 2 weeks ago (all delivered), but didn't send the ones that I ahdn't received the checks for
[07:55:23] <SWPLinux> K`zan: yes, but this is a different thing
[07:55:23] <A-L-P-H-A> oh
[07:55:24] <Jymmm> CBD
[07:55:24] <A-L-P-H-A> okay
[07:55:48] <SWPLinux> I got your check last Thursday or Friday, and didn't have a chance to deposit it and mail the box
[07:55:59] <SWPLinux> or maybe earlier - the days blur when you work at home :)
[07:56:18] <Jymmm> * Jymmm returns back to his CCNA reading
[07:56:19] <A-L-P-H-A> hmm... hope those boards will be fun... :)
[07:56:37] <K`zan> If I go with the pimo drivers I need to track down steppers and thrust bearings and ...
[07:57:00] <SWPLinux> they're pretty well made, and there are a few nice designs Peter did, that make me think they'll work very nicely
[07:57:06] <K`zan> Starting to like that idea as scary as it is :).
[07:57:18] <SWPLinux> what kind of power do you need?
[07:57:22] <SWPLinux> for the motors
[07:57:21] <K`zan> Me?
[07:57:24] <A-L-P-H-A> K`zan, are you going to move soon, so you can save pennies to get geckos?
[07:57:24] <SWPLinux> yep
[07:57:43] <K`zan> As near as I can tell, 269 0z/in should work well for the uMill.
[07:57:58] <SWPLinux> at what speed
[07:58:10] <Jymmm> of ligth
[07:58:11] <Jymmm> light
[07:58:18] <K`zan> SWPLinux: I really hate to admit this, but I have not a CLUE...
[07:58:19] <Jymmm> 186,000 m/s
[07:58:43] <K`zan> IIRC, the lead screws are 20tpi (I think).
[07:58:49] <SWPLinux> eek
[07:58:55] <K`zan> Stock...
[07:59:00] <SWPLinux> well, if you already have the steppers, then it's too late :)
[07:59:04] <ds3> whatkind of mill?
[07:59:08] <Jymmm> K`zan like a 1/4-20 bolt?
[07:59:10] <A-L-P-H-A> K`zan... www.roton.com get tiny ball screws. :D
[07:59:31] <K`zan> I don't, I've been thrashing mostly because of a budget from hell and a LOT to do to get it all together.
[08:00:05] <K`zan> Was going to throw money at it (which I really don't have) and get the xylotex 3 axis system and the CNC-Fusion mounts.
[08:00:39] <K`zan> I'm starting to be somwhat less terrified about making my own mounts...
[08:01:01] <SWPLinux> use bubble gum if it lets you afford better drivers
[08:01:02] <A-L-P-H-A> use thick rubber tubes, or just journals.
[08:01:07] <K`zan> Get something together and then improve on it.
[08:01:40] <A-L-P-H-A> xylotex is putting money into garbage... :) but it's your money... I really hope they'll be good enough for you.
[08:01:48] <K`zan> I was thinking of doing what I did with the floppy stepper thing - radiator hose and clamps - worked well but then again it was not driving anything.
[08:02:12] <A-L-P-H-A> K`zan, hose will work fine on a tiny machine like that.
[08:02:13] <K`zan> A-L-P-H-A: I have about decided that the xylotex suff is really delicate...
[08:02:20] <SWPLinux> Mariss said it well today: a stepper motor is only as good as the driver
[08:02:50] <Jymmm> SWPLinux and he sells drivers. what else would he say... he's a fgood salesman
[08:03:02] <K`zan> The ones I am thinking about (no idea which) are the boards from err, lemme get the link.
[08:03:05] <A-L-P-H-A> he is such a talker.
[08:03:19] <SWPLinux> he's a millionaire, and he sells 3000 drives/month to industry - do you think it matters much if you don't buy one?
[08:03:25] <K`zan> SWPLinux: sells drivers?
[08:03:29] <SWPLinux> geckos
[08:03:36] <SWPLinux> Mariss = geckodrive
[08:03:49] <SWPLinux> Mariss == geckodrive
[08:03:51] <K`zan> http://pminmo.com/
[08:03:56] <SWPLinux> retch
[08:03:59] <SWPLinux> sorry - gut reaction
[08:04:03] <K`zan> It will be a while before I can afford gekos
[08:04:06] <Jymmm> SWPLinux Hell yes... it's the salesman in him.
[08:04:27] <K`zan> $200 each is about what I paid for the mill, for each ones :-(.
[08:04:34] <SWPLinux> heh - I think I may know him a little better than that (I may not, but I think I do)
[08:04:47] <A-L-P-H-A> 3000/month x $75/pop whole sale... = 225000,,, say $25 profit each... 75000... and been doing this for 6 years... 5,400,000. He's not hurting.
[08:04:47] <K`zan> I am clueless here and working on an embarrassing budget....
[08:05:24] <SWPLinux> he used to be half owner of Centent, and he retired before deciding to start Geckodrive (because he wanted to pursue the ideas that he couldn't do at Centent)
[08:05:26] <Jymmm> K`zan Nah, no embarassing - seriosuly. You are just working within your means is all.
[08:05:26] <K`zan> I *can* build anything including SMT stuff.
[08:05:54] <SWPLinux> what motors were you thinking of?
[08:06:06] <SWPLinux> you mentioned 269 oz-in - there must be a specific one :)
[08:06:11] <ds3> K`zan: what kind of machine are you building?
[08:06:27] <K`zan> Jymmm: Well the budget is embarrasing :-), but I have no idea what I am doing and the more I research the confuzeder I get :-/.
[08:06:39] <SWPLinux> funny that pminmo links to Mariss' "step motor basics" document ;)
[08:06:45] <K`zan> SWPLinux: No idea now, I was going to get the xylotex kit...
[08:06:51] <SWPLinux> ah
[08:07:04] <SWPLinux> believe it or not, Automation Direct has pretty inexpensive steppers
[08:07:10] <K`zan> Not sure if getting those separately from him would be good / wise or not.
[08:07:16] <SWPLinux> like $29 or $39 each, for reasonable torque
[08:07:32] <K`zan> The ones from xylotex were $50 IIRC.
[08:07:36] <Jymmm> K`zan If I buy geckos, I'll sell you my xylotex board =)
[08:08:01] <K`zan> Jymmm: No offense intended, but after researching this for a while, would I want them?
[08:08:27] <Jymmm> K`zan Not on 20TPI, but on 5TPI sure
[08:08:31] <SWPLinux> http://web1.automationdirect.com/adc/Shopping/Catalog/Motion_Control/Stepper_Systems/Motors_-z-_Cables
[08:08:41] <SWPLinux> 276 oz-in, $39.00
[08:08:53] <K`zan> It seems they blow up on people regularly and people using them sure seem to have an inordinate amount of problems...
[08:08:58] <Jymmm> K`zan But, something to get started on is better than nothing at all.
[08:09:06] <SWPLinux> heh
[08:09:09] <K`zan> Jymmm: There IS that.
[08:09:14] <SWPLinux> plus it would help Jymmm get Geckos :)
[08:09:20] <Jymmm> K`zan Nah, you just have to be careful with your wiring is all
[08:09:22] <K`zan> I'm already real REAL tired of cranking on that thing :).
[08:09:43] <K`zan> Hard on arthritic wrists :-(.
[08:09:52] <K`zan> SWPLinux: Checking
[08:10:07] <SWPLinux> their drives are expensive, but the motors are reasonable
[08:10:13] <SWPLinux> unless you want AC servos
[08:10:30] <SWPLinux> I'd have to pay more for servos there than I did for my Bridgeport
[08:11:19] <K`zan> This is what I was looking at: http://www.xylotex.com/Econo3AxSpecial.htm
[08:12:01] <K`zan> SWP Unf I think the budget is going to force steppers :-(.
[08:12:09] <SWPLinux> yes
[08:12:20] <SWPLinux> what's the mill?
[08:12:34] <K`zan> Maybe when I makes lots of $$$ with it and can afford a bridgeport / clone :).
[08:13:01] <SWPLinux> heh - well, I got a pretty good deal, but then spent waaaay too much on parts (and still haven't made the thing into a CNC :( )
[08:13:22] <K`zan> SWPLinux: This one: http://www.harborfreight.com/cpi/ctaf/displayitem.taf?Itemnumber=47158
[08:13:54] <K`zan> I orderd a belt drive for it a while back and it should be here next week!!!
[08:13:56] <SWPLinux> you know, I wish people would undrestand that it's better to resize the image before sticking it on a web page, rather than having the browser squash it into oblivion with incorrect size tags
[08:14:01] <SWPLinux> ah, the HF minimill
[08:14:13] <SWPLinux> err -micromill
[08:14:14] <K`zan> Got tired of eating those damn plastic gears...
[08:14:18] <K`zan> SWPLinux: Yes
[08:14:22] <SWPLinux> ok. 269 oz-in should be plenty, I'd bet :)
[08:14:32] <SWPLinux> you can test that, by the way
[08:14:42] <Jymmm> They work on my 5TPI acme on a 2:1 gearing
[08:14:42] <K`zan> Those you pointed me at are 276 which is a bit more.
[08:14:50] <SWPLinux> if you can either estimate the force you have to put on the wheel, or if you can get a fisherman's scale
[08:15:15] <K`zan> As much as I hate to consider it, I think I might be wiser to get the long table mods for that thing first...
[08:15:33] <K`zan> Attach to handle crank and pull until it moves?
[08:15:38] <SWPLinux> yep
[08:15:48] <SWPLinux> or just push and see what it feels like
[08:15:55] <K`zan> Can probably get one from k-fart.
[08:16:01] <K`zan> mart <blush>
[08:16:05] <SWPLinux> the cranks are probably 3" diameter or something
[08:16:38] <SWPLinux> the xylotex motors are fairly high voltage for their size
[08:16:51] <K`zan> touch over 3"
[08:17:16] <K`zan> SWPLinux: is that good or bad?
[08:17:43] <SWPLinux> ok, so the radius is 1.6" maybe. unless you have to push with ~10 pounds of force, the 270-ish oz-in motors should be fine
[08:18:01] <SWPLinux> the voltage isn't great - 3.164V
[08:18:12] <K`zan> Biggest problem with that mill is that the Y axis is so small that by the time you get anything on the table, the Y is well into out of the gibs.
[08:18:19] <SWPLinux> you'd want to run at 60-80V to get the best performance, but the xylotx can't do that
[08:18:32] <SWPLinux> yes
[08:18:35] <K`zan> No, IIRC, 32V ABS MAX...
[08:18:50] <SWPLinux> yep
[08:18:51] <K`zan> After that the smoke escapes...
[08:19:08] <SWPLinux> there's current to spare I think - it can go to 5A or something (right jymmm?)
[08:19:25] <K`zan> I have a 24V 10A power wheelchair battery charger, dunno if that would work or not (butchering it is not a problem :-).
[08:19:29] <SWPLinux> eek - only 2.5A
[08:19:48] <K`zan> Big box :-).
[08:20:00] <K`zan> Big heavy transformer...
[08:20:15] <SWPLinux> it's not ideal
[08:20:39] <SWPLinux> hmmm - the AD motors are pretty close to the Xylotex ones in voltage and resistance
[08:20:46] <K`zan> May put out more with some mods to the output...
[08:21:05] <K`zan> Have to take it apart and play with it...
[08:22:23] <K`zan> I saw one web site where the guy used the stock attachment for the cranks to attach the steppers, I should be able to make the adapters here with what I have:
[08:22:47] <K`zan> http://www.theblossers.net/index.php?section=15
[08:22:59] <K`zan> See project files over on the middle right...
[08:23:38] <K`zan> If I can get it together at all, I can start improving it FAR more easily.
[08:24:28] <SWPLinux> nice flat mounts
[08:24:30] <SWPLinux> sigh
[08:24:44] <Jymmm> SWPLinux: 3A
[08:24:57] <SWPLinux> theysay 2.5 on the webiste, it seems
[08:25:03] <K`zan> Far from idea from what I gather, but I really don't want to spend money on a rotary table or a 4 jaw chuck just to do that.
[08:25:17] <K`zan> That is what CNC is for :).
[08:25:19] <Jymmm> SWPLinux: I have my PS tuened up to 27Volts per Jef''s ok
[08:25:22] <Jymmm> turned
[08:25:34] <SWPLinux> I'm just going from this: http://www.xylotex.com/3axBoard.htm
[08:25:39] <K`zan> Coffee, brb, freezing for some stupid reason...
[08:25:59] <Jymmm> SWPLinux looking...
[08:26:05] <SWPLinux> they should add emc to the list of software links
[08:26:50] <SWPLinux> and the link they actually have to emc is stale
[08:26:53] <Jymmm> SWPLinux Note the +-2.5 A/phase
[08:27:01] <SWPLinux> yes, I did
[08:27:17] <SWPLinux> 2.5 != 3
[08:27:19] <Jymmm> SWPLinux I have my XY at 3A, and my Z at 2A
[08:27:43] <SWPLinux> so Z will continue to work, and X/Y will blow up at some point - simple ;)
[08:28:13] <Jymmm> SWPLinux If it does, it's per Jeff's instructions.
[08:28:23] <SWPLinux> Jeff@xylotex?
[08:28:28] <Jymmm> yep
[08:29:12] <Jymmm> There really is no problem with the board, it's just that geckos have mid band compensation is all, and that's what I need.
[08:30:16] <Jymmm> Now, I do have issues on my setup after 90+ minutes in stalling. So, I now have (just this week) an Non Contact thermometer and I'll be able to see if the motor are getting hot and if it's a thermal issue or something else.
[08:30:36] <Jymmm> I do have like 7 fans on my PS/driver, so I dont' THINK that's the issue there.
[08:30:39] <SWPLinux> you mean after 90 minutes stalled (holding)?
[08:31:01] <Jymmm> SWPLinux steppers stall for a moment, then ruin the piece being run.
[08:31:15] <SWPLinux> ah - you get stalls after 90+ minutes of operation ...
[08:31:45] <Jymmm> 90 to 120 minutes. Never been constantly been able to repeat it; thus harder to find the cause.
[08:32:25] <SWPLinux> if possible, try aiming a fan at the motors
[08:32:40] <SWPLinux> is it usually one axis, or random?
[08:32:47] <Jymmm> But, had a long conversation with MAriss and geckos have mid band compensation, so they should resolve the problem.
[08:32:54] <Jymmm> random, x or Y
[08:33:00] <SWPLinux> depends on the problem
[08:33:07] <SWPLinux> never Z?
[08:33:22] <Jymmm> SWPLinux That's one reason why I bought the IR Thermometer - instant reading.
[08:33:55] <Jymmm> SWPLinux Nope, but Z doesn't do much anyway..... max travel during a run is MAYBE 1.5"
[08:34:12] <SWPLinux> holding is also work
[08:34:21] <SWPLinux> well, effort, not work
[08:34:41] <Jymmm> not really, 5TPI direct
[08:35:03] <SWPLinux> do you use router bits or spiral flute bits?
[08:35:20] <Jymmm> Yes =)
[08:35:28] <SWPLinux> actually, it doesn't matter with steppers
[08:35:44] <SWPLinux> the driver still pumps full current (only 2A) through the motor to keep it in place
[08:36:04] <SWPLinux> and it actually should get warmer, since the rotor isn't moving (hence no internal airflow)
[08:36:45] <Jymmm> The Z is on more aluminum than Xa nd Y are,
[08:36:58] <Jymmm> (heatsink)
[08:37:28] <SWPLinux> it also gets <1/2 the heating due to the lower current
[08:37:36] <SWPLinux> (square law)
[08:37:55] <Jymmm> BUT.... I had the whole machine in an dust proof eonclosure that also kept the heat in too. Wehn I setup this time, I've already workked on using canvas to the heat can escape better.
[08:38:49] <SWPLinux> you have fans aimed at the driver board?
[08:39:01] <Jymmm> 7
[08:39:27] <SWPLinux> is the driver mounted on/near the PS?
[08:39:33] <K`zan> Decided on a LARGE mug of hot chocolate :).
[08:39:34] <Jymmm> yep
[08:40:10] <SWPLinux> ok. was that on or near? :)
[08:40:29] <Jymmm> lol.... near, about 4" away parallel to each other
[08:40:49] <SWPLinux> hmmm. that sounds like enough of a wind tunnel
[08:41:11] <SWPLinux> I'd try setting the current back to 2.5A or less (even though you;ll need to slow the machine down a bit), and see what happens
[08:41:29] <Jymmm> SWPLinux how come?
[08:41:43] <K`zan> SWPLinux: Am I correct that you are a geko dealer?
[08:41:49] <SWPLinux> nope
[08:42:03] <SWPLinux> I bought a few servo drives from them, and I may even use them :)
[08:42:11] <SWPLinux> Jymmm: heat
[08:42:36] <K`zan> Err, then I evidentally missed something there, I got that you sold drives and somehow gekkos came into it.
[08:42:40] <Jymmm> SWPLinux WEll, I can give it a shot. easy enough to try.
[08:42:41] <SWPLinux> the manual for the drive really stresses the 2.5A limit, and recommends a heatsink and cooling when operating near the 2.5A limit
[08:42:56] <Jymmm> SWPLinux I really do have a LOT of fans on it.
[08:43:12] <Jymmm> SWPLinux 4"
[08:43:13] <SWPLinux> K`zan: no problem. I recommend the geckos since I know people with them, and they're inexpensive for the performance you get from them
[08:44:02] <SWPLinux> I also like the attitude of the company (their return policy is very good)
[08:44:57] <SWPLinux> ok. it looks like the xylotex doesn't have idle current reduction either
[08:45:15] <SWPLinux> (which is good when you're trying to hold Z steady for long periods, I guess)
[08:45:41] <Jymmm> well, it sorta does, but it's manual
[08:45:44] <K`zan> <K`zan> SWPLinux: sells drivers?
[08:45:44] <K`zan> <SWPLinux> geckos
[08:45:48] <Jymmm> you have to flip a switch
[08:46:04] <K`zan> I do manage to stay confused with all this :).
[08:46:18] <K`zan> ah, uh.... :-)
[08:46:25] <Jymmm> SWPLinux there's a header ont eh board that allows to you to places each axis into "standby"
[08:46:40] <SWPLinux> heh
[08:46:50] <SWPLinux> K`zan: right - jymmmmmmm was referring to Mariss ... :)
[08:47:04] <Jymmm> SWPLinux no xylotex
[08:47:08] <SWPLinux> them too
[08:47:16] <SWPLinux> in the quoted text
[08:47:18] <SWPLinux> or just before it
[08:47:21] <K`zan> err Mariss == gekos somewhere back up there...
[08:48:00] <K`zan> Mariss is a name not a product, OK...
[08:48:10] <K`zan> Sigh....
[08:48:28] <SWPLinux> Mariss Freimanis is the owner of Geckodrive, Inc
[08:48:41] <SWPLinux> he's also the designer of all their products
[08:49:01] <K`zan> Ah, I was starting to get the idea you were into the slave trade selling Mariss's :-).,
[08:49:11] <K`zan> LOL
[08:49:24] <K`zan> Yes, it has been a long day, why do you ask :)?
[08:49:49] <SWPLinux> no reason. oh wait - maybe it's because it's almost 4AM here
[08:50:02] <K`zan> 0049 here :-).
[08:50:32] <Jymmm> K`zan what the hell?
[08:50:37] <Jymmm> oh, nm =)
[08:50:40] <K`zan> ??
[08:50:53] <K`zan> Don't catch my confuzion :)
[08:50:56] <SWPLinux> err - jymmm - it's the same time as you're in ...
[08:51:21] <Jymmm> SWPLinux Yeah, but stupid clock here is in 12h formnat
[08:51:27] <SWPLinux> heh
[08:51:33] <SWPLinux> stupid americans
[08:51:39] <K`zan> Hate those, never know if it is morning or evening...\
[08:51:40] <Jymmm> not used to 12h format, only 24h
[08:51:46] <K`zan> Same here.
[08:52:15] <K`zan> ALL my clocks with the exception of the TV and VCR/DVD are 24 hour or out they go :-).
[08:52:45] <SWPLinux> well, 3:52 is the same no matter which scheme you use
[08:52:47] <SWPLinux> good night :)
[08:52:53] <K`zan> LOL true...
[08:52:53] <Jymmm> Yep, took a nap one time, woke up look at the clock it was 6:30am I go oh shit... then I realize it just got dark early and it was 1830 not 0630 and bought a 24h clock the next day
[08:53:12] <K`zan> Night, rest well and thanks for the help in de-confusing, I'll get there :)
[08:53:23] <SWPLinux> have fun - remember it's the journey
[08:53:30] <SWPLinux> or some such claptrap
[08:53:35] <K`zan> Yep, it is fun and will get better :)
[08:55:37] <Jymmm> My newest toy... http://www.harborfreightusa.com/usa/itemdisplay/displayItem.do?itemid=90714&CategoryName=&SubCategoryName=
[08:56:14] <K`zan> about $150 for the steppers and the 20' extension cables.
[08:56:50] <K`zan> Looks good, good quality?
[08:56:51] <Jymmm> SWPLinux: Thsi si the one I got... http://www.harborfreight.com/cpi/ctaf/displayitem.taf?Itemnumber=91778
[08:59:33] <K`zan> Looked at that, let me know how it works out?
[09:00:12] <Jymmm> K`zan Not bad at all actually. Operation deisgned well. I haven't actually tested it's accuracy yet, but seems pretty much on the nose.
[09:00:43] <K`zan> Relative should be OK, but good to know you life it :)
[09:01:27] <Jymmm> I shot it at the oven set to 350, read as 354
[09:02:02] <Jymmm> will have to boil some water and take a reading.
[09:02:10] <K`zan> Knowing how oven termos work, I;d trust the meter :).
[09:02:35] <Jymmm> heh, and this oven is like 20 years old too
[09:03:07] <K`zan> Not bad.
[09:03:20] <K`zan> One here is going on 40 :).
[09:04:06] <Jymmm> I like that it auto shuts off after 7 seconds or so. And when you let go of the trigger it HOLDS the reading on the display for you. NICE backlight and laser too
[09:04:20] <Jymmm> one 9v battery
[09:04:29] <K`zan> Yep, looked at that and almost got it....
[09:04:51] <Jymmm> I've been wanting one for years. Saw it for $40 and had to try it out at leas.t
[09:04:56] <K`zan> When $$$ free up some, I'll go ahrad and get it.
[09:05:05] <K`zan> Price is excellent.
[09:05:32] <K`zan> Problem being my age, a lot of stuff I really lusted after over the years is now dirt cheap :).
[09:05:55] <Jymmm> It's hard to justify a theremometer, but if it helps resolve issues with my machine, It'll pay for itself in tequilia and asprin.
[09:06:27] <K`zan> Well, time to splat, LONG day. Catch ya on the morrow. Thanks for the info and help and wisdom and all that. Rest well.
[09:06:49] <Jymmm> Eh, I'm still at work for another couple hours
[09:07:15] <K`zan> Understand, I got the clamp meter so the landlord would quit foaming at the mouth but now it is easy to check things for peace of mind :).
[09:07:23] <K`zan> Where are you?
[09:07:35] <Jymmm> Northern California
[09:07:47] <K`zan> Ah, same TZ I am.
[09:07:51] <Jymmm> yep
[09:07:51] <K`zan> Nice shift :-).
[09:08:25] <K`zan> After working 2nd and 3rd shift for years, the only way I sleep at night is if I am exhausted.
[09:08:26] <Jymmm> K`zan Not bad, I still get part of the day to myself,
[09:08:36] <K`zan> THat is the nice part of second.
[09:08:40] <Jymmm> K`zan: I go stright to sleep when I get home
[09:08:49] <Jymmm> K`zan 1900-0300
[09:08:57] <Jymmm> home by 0330
[09:09:06] <K`zan> I'm just getting ready to go to work when most people are ready to go to bed :).
[09:09:09] <Jymmm> err in bed (literally) by 0330
[09:09:15] <Jymmm> lol
[09:09:40] <K`zan> Anyhow, time to fall over, catch ya later.
[09:09:46] <Jymmm> G'Night
[10:10:14] <lerneaen_hydra> 'mornin everyone
[10:12:27] <A-L-P-H-A> hey
[10:13:42] <alex_joni> morning lerneaen_hydra
[10:13:44] <alex_joni> hi A-L-P-H-A
[10:13:45] <A-L-P-H-A> hey hey
[10:14:03] <A-L-P-H-A> my ears feel so dirty. :(
[10:14:49] <A-L-P-H-A> q-tips gets so much wax is so gross... I should go to the doctor about my ears...
[10:25:57] <lerneaen_hydra> nasty
[10:28:01] <A-L-P-H-A> cotton swab made my ears feel cleaner.
[11:00:44] <A-L-P-H-A> it's hailing outside.
[11:00:47] <A-L-P-H-A> wow...
[11:00:54] <A-L-P-H-A> or little freezing rain...
[11:01:01] <A-L-P-H-A> difference, I don't really know.
[11:01:31] <lerneaen_hydra> here it's just cloudy
[11:02:23] <A-L-P-H-A> lerneaen_hydra, what's new? more blog posts?
[11:02:29] <lerneaen_hydra> nah, not much
[11:02:37] <lerneaen_hydra> not much at all
[11:02:59] <lerneaen_hydra> lots of stuff I should be doing though
[11:04:23] <A-L-P-H-A> such as?
[11:04:33] <A-L-P-H-A> I should really call that girl from Saturday night today.
[11:14:11] <lerneaen_hydra> such as that for instance
[11:20:03] <Martin_Lundstrom> go and get them lerneaen_hydra :)
[11:21:37] <alex_joni> wonder how the chicks dig 7 heads
[11:21:49] <Martin_Lundstrom> lol
[11:22:40] <lerneaen_hydra> hahaha
[11:22:53] <Martin_Lundstrom> lerneaen_hydra, maybe you can try some concieling makeup for the other 6
[11:22:58] <alex_joni> 5
[11:23:05] <alex_joni> 2 thez will need
[11:23:09] <alex_joni> S+P
[11:23:19] <Martin_Lundstrom> are u dirty now?
[11:23:22] <alex_joni> darn language & kezboard
[11:23:38] <lerneaen_hydra> I think alex sure seems to be at least
[11:24:16] <Martin_Lundstrom> lerneaen_hydra, you know these romanian people...
[11:24:24] <lerneaen_hydra> yeah
[11:24:32] <alex_joni> synergy has problems with 2 pcs using different languages
[11:24:48] <lerneaen_hydra> we know all about them from borat, which was exceedingly good at depicting reality
[11:25:33] <Martin_Lundstrom> is not borat from kazakstan?
[11:25:39] <alex_joni> lerneaen_hydra: you're lucky I'm not ill tempered now :P
[11:25:53] <alex_joni> Martin_Lundstrom: yeah, but it was filmed in romania :P
[11:25:55] <lerneaen_hydra> or else what? :p
[11:26:00] <lerneaen_hydra> * lerneaen_hydra what alex said
[11:26:10] <alex_joni> alex_joni has kicked lerneaen_hydra from #emc
[11:26:22] <alex_joni> or else .. nothing
[11:26:24] <lerneaen_hydra> boo hoo hoo
[11:26:43] <lerneaen_hydra> why do I always forget that alex can get +o here
[11:26:50] <Martin_Lundstrom> alex_joni, is it fun to play god?
[11:26:54] <lerneaen_hydra> that reminds me of a bash quote
[11:27:00] <alex_joni> Martin_Lundstrom: not really
[11:28:41] <lerneaen_hydra> this one! :D http://www.bash.org/?2605
[11:29:29] <lerneaen_hydra> damn, is it a bad thing that I remember most of the bash quotes in the top 200 list?
[11:29:40] <alex_joni> no, I do that too
[11:29:53] <Martin_Lundstrom> thats amusing
[11:30:22] <alex_joni> http://bash.org/?33
[11:30:58] <lerneaen_hydra> haha
[11:31:19] <lerneaen_hydra> hmm, bbl
[11:32:42] <alex_joni> /quit and /exit were sitting in a park. /exit left, so who was still sitting in the park?
[11:33:21] <Jymmmm> Jymmmm has kicked alex_joni from #emc
[11:33:23] <Martin_Lundstrom> /quit ;)
[11:33:45] <alex_joni> funny :P
[11:33:57] <alex_joni> http://bash.org/?447880
[11:49:19] <alex_joni> http://bash.org/?177638 rofl
[11:55:57] <A-L-P-H-A> alex_joni, that last one uber!
[11:56:03] <A-L-P-H-A> ahhhh...
[11:59:40] <alex_joni> http://bash.org/?183544 lol
[11:59:56] <A-L-P-H-A> lol
[12:04:13] <A-L-P-H-A> this is just so random. http://bash.org/?636661
[12:04:18] <A-L-P-H-A> it's funny to me
[12:07:46] <alex_joni> some sick things on bash.org
[12:07:54] <alex_joni> http://bash.org/?261931
[12:08:00] <A-L-P-H-A> sick? http://bash.org/?22170
[12:09:18] <A-L-P-H-A> alex_joni, OMG. LOL
[12:09:19] <A-L-P-H-A> sick.............
[12:09:26] <A-L-P-H-A> SICK but soooo funny
[12:10:16] <Guest924> hello. I can't get ubuntu ready to run on compaq 800 mhz, onboard video (is this a problem?)
[12:10:37] <A-L-P-H-A> Guest924, ubuntu forums, search for your compaq model.
[12:10:38] <alex_joni> Guest924: unless you really want to use it... no
[12:11:00] <A-L-P-H-A> onboard video usually causes a problem... I had that with my friends compaq P3 onboard vid...
[12:11:10] <Guest924> Computer hangs during ubuntu screen: loading root file system
[12:11:29] <Guest924> okay, so i have to look for a pci video, and disable onboard?
[12:11:43] <A-L-P-H-A> Guest924, that's what I had to do... but I just did this instead, not load X.
[12:11:53] <A-L-P-H-A> and use it as a LAMP server.
[12:11:56] <A-L-P-H-A> not emc.
[12:12:10] <Guest924> have to say that i'm a total noob with linux this far.
[12:12:20] <A-L-P-H-A> then again, the point of that machine wasn't to play with emc, but use it as a torrent box
[12:12:36] <Guest924> okay.
[12:12:38] <A-L-P-H-A> alex_joni is a guru... you should message him about everything. :)
[12:12:47] <Guest924> thanks
[12:12:53] <Guest924> will message him
[12:15:04] <A-L-P-H-A> http://bash.org/?81904 so stupid.
[12:15:32] <A-L-P-H-A> http://bash.org/?114347
[12:16:37] <A-L-P-H-A> http://bash.org/?553790
[12:20:16] <A-L-P-H-A> http://bash.org/?18 hahaha
[12:20:39] <ottos> g d gents
[12:20:50] <A-L-P-H-A> aussie aussie aussie
[12:22:23] <A-L-P-H-A> http://bash.org/?713428 hahah
[12:25:24] <A-L-P-H-A> http://bash.org/?714249 ahahaha.
[12:25:28] <Guest991> Is there anyway to install ubuntu livecd without doing graphics during install
[12:26:00] <A-L-P-H-A> http://bash.org/?714336
[12:26:10] <A-L-P-H-A> Guest924, use... ubuntu alternative install cd.
[12:26:22] <A-L-P-H-A> it'll come up with a text-based install...
[12:26:58] <A-L-P-H-A> Guest924, it actually works, the graphics card, but just slow as hell... find a pci card... or get a used P4... they shouldn't be too expensive at a local flee market.
[12:28:05] <Guest991> okay.. thanks for the answers
[12:29:25] <A-L-P-H-A> Guest924, it's an intel 820i card, most likely... which means it shares it's video memory with your ram... kinda a cheap way to make computers back then... if you have small amount of ram, it sucks... as it's trying to do too much that the card can't handle...
[12:29:49] <Guest991> its having 256 mb ram
[12:29:58] <Guest991> but the video is sucking 64mb of that
[12:30:29] <alex_joni> Guest991: then you need more at least during install
[12:30:30] <rafa> rafa
[12:30:49] <rafa> hello friends!
[12:30:57] <Guest991> really? I thought it was more then enough
[12:31:12] <rafa> watt is ferror?
[12:31:14] <alex_joni> 256 is needed
[12:31:26] <Guest991> so.. guess thats causing the problem then
[12:31:29] <ottos> well installed a new version with all the updates.. I'm wondering if someone can take a peek at my ini files and make any suggestions..?
[12:31:32] <rafa> watt is ferror?
[12:31:34] <A-L-P-H-A> Guest991, you could install the textversion, and then sudo apt-get install ubuntu-desktop, which will install the desktop aftewrwards... that's one method.
[12:32:08] <rafa> watt is ferror?
[12:32:23] <Guest991> okay, but i have to get the alternative version to install text-based.. can't do with ubuntu live cd right?
[12:32:39] <A-L-P-H-A> I think ferror, is a following error... but I'm probably wrong.
[12:33:00] <rafa> how it function?
[12:33:00] <A-L-P-H-A> Guest991, #ubuntu :) would be the best choice to get help.
[12:33:27] <Guest991> thank you.. will go there.. youve been a great help.. :)
[12:33:43] <A-L-P-H-A> rafa, it's probaly your ini file that's causing the problem... but I'm not an expert. alex_joni would know best.
[12:33:45] <A-L-P-H-A> alex_joni... :)
[12:33:57] <A-L-P-H-A> but I think he's reading bash.org
[12:34:18] <alex_joni> phone
[12:34:50] <alex_joni> rafa: ferror means following error
[12:34:59] <alex_joni> that means your axis isn't moving as fast as it is commanded
[12:35:19] <alex_joni> if you are running stepper motors it means that you commanded a speed the realtime system can't acheive
[12:37:47] <A-L-P-H-A> alex_joni. hahahhahaha. http://bash.org/?715642
[12:38:18] <alex_joni> :-)
[12:40:04] <A-L-P-H-A> this made me lol. http://bash.org/?707531
[12:40:49] <alex_joni> http://bash.org/?50576
[12:41:14] <A-L-P-H-A> lol
[12:42:03] <A-L-P-H-A> hahaha. http://bash.org/?707847 I love the logic in that arguement.
[12:43:45] <ottos> alex.. any suppestion on how to improove the block process without killing the system..?
[12:44:14] <alex_joni> block process?
[12:44:17] <A-L-P-H-A> the base period?
[12:44:22] <ottos> yes..
[12:44:38] <alex_joni> ottos: thought you were runinng a servo ?
[12:44:49] <A-L-P-H-A> ottos, don't run other stuff. get a better computer... go with a mesa PCI card, when jmkasunich makes it for step and direction :)
[12:45:08] <A-L-P-H-A> I think he's running steppers. :D
[12:45:11] <A-L-P-H-A> that's my guess...
[12:45:12] <ottos> I am.. still does it
[12:45:46] <alex_joni> A-L-P-H-A: he's not using steppers
[12:45:57] <alex_joni> A-L-P-H-A: motenc
[12:46:02] <A-L-P-H-A> okay
[12:46:04] <ottos> yaskawa servopack in torque mode.. sdga.. motenc card..
[12:46:09] <A-L-P-H-A> I LOVE AMERICANTS.... http://bash.org/?709974
[12:46:46] <alex_joni> ottos: then you shouldn't care about base_period
[12:47:24] <ottos> what param will improove the traj performance..? servo and traj params..?
[12:47:57] <ottos> I've managed to lower them to 125000 and 250000 with base of 25000 but no smooth ride.. still jerks
[12:48:30] <A-L-P-H-A> gonna lay down.
[12:48:31] <A-L-P-H-A> ciao
[12:48:58] <alex_joni> ottos: are you sure it's not related to tuning?
[12:49:32] <alex_joni> ottos: can't really say what to check.. but I would start by scoping velocity and acceleration
[12:49:38] <alex_joni> halscope helps a lot there
[12:49:52] <alex_joni> then check if accel exceed the value you have in the ini..
[12:52:17] <ottos> all of the pid and other settings stayed the same.. I'll take a look at them again...
[12:52:34] <ottos> my max vel is 140mm max acceleration is 450 reg, 600 max..
[12:52:43] <alex_joni> I dont know what you mean by jerky
[12:52:51] <alex_joni> reg?
[12:53:53] <ottos> say you have a pattern and the motion should be fluid, contouring for example.. when hitting a straight line machine speeds up and goes ok, in the curves etc is slows almost to stop and them accelerates..
[12:54:23] <awallin> ottos, are you sure you are in G64 mode?
[12:54:35] <ottos> let me check..
[12:56:54] <awallin> G64Px.xx will specify a tolerance, so for example G64P0.01 to follow the programmed path with 0.01 tolerance
[12:58:40] <ottos> let me try with the g64 params..
[12:59:08] <alex_joni> morning anders
[12:59:33] <awallin> hi alex... 15 oclock around here, but maybe that's still morning ;)
[12:59:43] <alex_joni> same here :)
[13:00:09] <alex_joni> thought you were GMT+1 ?
[13:00:41] <awallin> no, +2 normally, don't what summer/winter time does to that
[13:00:53] <awallin> swe,nor,den are all +1
[13:01:05] <alex_joni> we are GMT+2+EEST during summer
[13:01:10] <alex_joni> GMT+2 now
[13:02:04] <alex_joni> * alex_joni wonders if 24 pages is a bit much for a quotation
[13:02:37] <awallin> just give us the first three lines...
[13:02:41] <awallin> if it's interesting
[13:02:53] <alex_joni> first 2 pages are yaddayadda
[13:07:14] <awallin> alex_joni: so what is it you are reading??
[13:07:46] <alex_joni> writing
[13:07:49] <alex_joni> :(
[13:09:15] <alex_joni> I have 3 quotations for today.. all about that long :/
[13:09:25] <ottos> ok .. getting better .. hope so..
[13:09:42] <alex_joni> ottos: you can set even lower precision to get higher speed
[13:09:45] <alex_joni> G64P0.1
[13:09:56] <alex_joni> but you know what precision you need
[13:10:15] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TrajectoryControl
[13:10:19] <alex_joni> it's described there
[13:11:05] <alex_joni> you can even use G64 without a P, that way the TP will blend as much as possible to keep up with the speed
[13:11:26] <awallin> alex_joni: define "as much as possible"
[13:11:42] <alex_joni> awallin: still needs to touch the single lines in at least one point
[13:12:13] <alex_joni> http://wiki.linuxcnc.org/uploads/g64-3-small.png <- kinda like the worst case in that pic
[13:13:45] <awallin> alex_joni: I think only G64 without P is a risky way of programming
[13:14:00] <awallin> won't produce the same part on different machines
[13:14:42] <ottos> slex do you think 70 block a sec is a overkill?
[13:14:54] <ottos> alex..sorry..
[13:18:01] <alex_joni> what is a block ?
[13:18:19] <ottos> 70 block lines a sec.. for the traj to handle..
[13:18:51] <alex_joni> you mean 70 lines from the file?
[13:19:05] <alex_joni> lines/instructions/whatever?
[13:19:39] <ottos> yes.. Nxx lines..
[13:20:26] <alex_joni> should be OK
[13:24:04] <alex_joni> anzone knows "Jerry Goldsmith - Life is a dream" ?
[13:27:47] <ottos> alex I've tried the g64 param.. it helps a bit but the performance of the machine is directly related to the number of line s/ sec it has to process. I've created a one with 10 blocks/sec and it's not too bad..
[13:31:44] <alex_joni> ottos: I see..
[13:32:04] <alex_joni> maybe you can convince your CAM to output G2/G3's instead of small line segments?
[13:32:07] <jepler> ottos: are the individual motions long enough to have a "cruise phase" with your machine's acceleration and the requested feed rate?
[13:32:32] <alex_joni> oh, hi jeff.. just the one to ask about segment throughput :P
[13:32:40] <jepler> alex_joni: no, that's not me -- that's cradek
[13:32:53] <alex_joni> :-)
[13:32:58] <ottos> I'm trying to test that right now.. might word..
[13:33:02] <jepler> unless it's about the changes I made to the userspace side
[13:33:08] <alex_joni> I know you looked at the throughput at some point..
[13:33:11] <alex_joni> oh, right
[13:34:50] <jepler> the absolute maximum transfer rate of segments from userspace to realtime is the "servo period", typically 1ms. because it involves a very simplistic method of handshaking with userspace, the actual rate is somewhat lower.
[13:35:22] <alex_joni> yes, ottos reported his results got better by reducing the traj/servo periods
[13:36:09] <jepler> I just performed a test, putting the length of the realtime trajectory queue in location I could show in halscope and setting feed override to 0%
[13:36:26] <jepler> the actual rate I got was about 100/second
[13:36:32] <ottos> yes.. I cannot explain why but on the old system 5.10 and 2.0.1 with base of 20000 traj at 200000 and servo at 100000 emc2 was flying..
[13:37:17] <jepler> I'm not sure if 2.0.x has the improvements I made in the transfer of segments from userspace to realtime
[13:37:23] <jepler> I suspect it doesn't
[13:37:54] <jepler> before my changes, the userspace task did a "sleep" no matter whether it transferred a trajectory item, giving lower throughput from userspace to realtime
[13:38:29] <ottos> jepler how do I check for the queue lenght in halscope.. hal..?
[13:38:32] <jepler> so unless you're using 2.1, my remarks probably don't apply
[13:38:52] <jepler> ottos: I think you have to modify the source code to be able to halscope this particular value
[13:39:59] <ottos> ok.. I'll skip on that one..
[13:40:06] <A-L-P-H-A> can't sleep
[13:40:09] <alex_joni> ottos: in a couple of days I hope there will be some pretesting package for 2.1.0
[13:40:24] <alex_joni> ottos: it would be appreciated if you could try it out
[13:40:38] <jepler> http://pastebin.ca/317286
[13:40:50] <ottos> no prob at all.. woudl be nice to have my cnc working 100%
[13:41:25] <jepler> with this patch applied, I halscope motion.debug-float-1
[13:41:44] <alex_joni> jepler: 2 line patch, but 20 minutes to set up the compile :)
[13:41:55] <jepler> alex_joni: yes that may be true
[13:42:13] <ottos> trying arc etc.. one sec..
[13:42:26] <alex_joni> I assume a semi advanced user and a somehow decent connection
[13:46:28] <CIA-8> 03jepler 07TRUNK * 10emc2/docs/src/hal/pluto_servo.html: remove note about CVS version required, since any version with this file is appropriate
[13:49:36] <CIA-8> 03jepler 07v2_1_branch * 10emc2/docs/src/ (Submakefile index.tmpl): merge pluto documentation from HEAD
[13:49:38] <CIA-8> 03jepler 07v2_1_branch * 10emc2/docs/src/hal/pluto_servo.html: merge pluto documentation from HEAD
[13:50:54] <ottos> arc setup help.. linear segments still jerk a bit..
[13:55:11] <ottos> arc helps but still not 100%
[13:57:28] <jepler> I hate to ask the obvious here, but if 2.0.1 works better, have you tried downgrading to that version? sudo apt-get install emc2=1:2.0.1
[13:58:13] <jepler> if you can verify that with everything else the same, 2.0.1 "works better than" 2.0.5, then the developers can treat it like they would any bug that was introduced
[14:00:16] <ottos> no .. kinda rusty with linux.. :D
[14:00:34] <ottos> let me try..
[14:00:44] <jepler> (I got the impression you changed several things at around the same time -- configuration, emc version, and ubuntu version)
[14:01:17] <ottos> the config is the same.. I mean the params.. all other ver are changed..
[14:01:38] <alex_joni> ottos: thats not the best waz to hunt an error down
[14:01:43] <alex_joni> way even
[14:03:38] <ottos> jepler If I install 2.0.1 how do I get back to 2.0.5 in case..?!
[14:06:13] <jepler> ottos: this will return you to the latest version: sudo apt-get install emc2
[14:12:38] <CIA-8> 03jepler 07v2_1_branch * 10emc2/debian/ (emc2-docs.files rules.in): include pluto html documentation in the -docs package
[14:13:26] <CIA-8> 03jepler 07TRUNK * 10emc2/debian/ (emc2-docs.files rules.in): include pluto html documentation in the -docs package
[14:14:37] <vq> vq is now known as Vq^
[14:15:20] <jepler> How long has that message said "TRUNK"?
[14:15:22] <ottos> ok 2.0.1 on 6.0.6 has a timing diffs.. with .05 the same values almost lock the system with .1 it still workable..
[14:16:17] <jepler> is the hardware identical?
[14:16:43] <ottos> yes..
[14:17:59] <jepler> unfortunately, then, we've only narrowed it down to all the difference between the two major versions of ubuntu linux
[14:18:01] <alex_joni> jepler: just started
[14:18:25] <awallin> so now, TRUNK==HEAD ?
[14:18:57] <alex_joni> Branch: TRUNK
[14:19:34] <ottos> hmm.. not helpfull is it...
[14:20:33] <alex_joni> ottos: what kind of pc is this?
[14:21:36] <ottos> p4 1.8Ghz, 512Mb ram..
[14:22:24] <ottos> I used to run a p3 with the same setup and it same the same old.. I've upgraded to the p4 and lowered the values and finally it worked.. up to yesterday..
[14:23:51] <jepler> what did you change yesterday?
[14:24:58] <ottos> had to reload the sytem from what ever reason emc was giving me out of limit errs on anythign I've done..
[14:25:33] <jepler> oh -- I remember everyone here advising that you not do that
[14:26:07] <jepler> I'm sorry -- I shouldn't have said that. it's not helpful at this point.
[14:26:26] <ottos> was the only resource.. system was unusable..
[14:26:49] <jepler> before yesterday were you using ubuntu 5.10 "breezy"?
[14:26:50] <ottos> no mdi or auto..only jog..
[14:26:53] <ottos> yes..
[14:28:54] <jepler> http://old-releases.ubuntu.com/releases/5.10/ubuntu-5.10-install-i386.iso
[14:29:02] <jepler> you could try installing 5.10 again
[14:29:06] <awallin> does 6.06 vs. 6.10 make any difference?
[14:29:09] <awallin> for emc
[14:29:24] <jepler> awallin: we don't make a kernel or an emc2 package for 6.10
[14:30:05] <alex_joni> you can run sim on 6.10 though :)
[14:30:16] <alex_joni> and it will run faster than a stepper :P
[14:30:34] <skunkworks> attos: how did you ubgrade to dapper?
[14:30:51] <ottos> iso install...
[14:30:56] <skunkworks> ok
[14:31:47] <ottos> d/l the old ver , will test it.. is the .sh for the breeze still usable..?
[14:32:04] <alex_joni> ottos: yes
[14:35:21] <lerneaen_hydra> iab!
[14:36:49] <jepler> ugh -- work is calling
[14:36:50] <jepler> bbl
[14:40:05] <lerneaen_hydra> hahaha
[14:40:16] <lerneaen_hydra> such exquisite timing
[14:41:16] <alex_joni> ??
[14:41:40] <lerneaen_hydra> I come back, jepler has to go to work :D
[14:41:46] <lerneaen_hydra> "go to work" ;)
[14:42:00] <lerneaen_hydra> that reminds me of a bash quote
[14:42:20] <alex_joni> of course it does
[14:42:22] <alex_joni> :/
[14:43:00] <lerneaen_hydra> most things remind me of bash quotes
[14:47:12] <lerneaen_hydra> hmm, can't find it
[14:47:14] <lerneaen_hydra> damn
[15:15:35] <lerneaen_hydra> so... what's happening?
[15:27:47] <alex_joni> (Connection reset by peer) ?
[15:27:56] <alex_joni> usually it means that's a bad conenction :)
[15:31:47] <tomp> awallin: i'm restarting on the widget, chucking old idea from tkdial.tcl
[15:33:00] <tomp> awallin: in emc-devel
[15:34:22] <lerneaen_hydra> alex_joni: would that be 2.1 testing?
[15:34:31] <lerneaen_hydra> s/2.1/2.1.x
[15:34:39] <alex_joni> lerneaen_hydra: would what be?
[15:35:12] <lerneaen_hydra> the testing package that the board had decided on
[15:35:22] <alex_joni> not really decided
[15:35:23] <lerneaen_hydra> but nobody knew the conclusion :p
[15:35:25] <alex_joni> :D
[15:35:27] <lerneaen_hydra> in emc-devel
[15:35:29] <alex_joni> yeah
[15:35:50] <lerneaen_hydra> ok
[15:55:59] <lerman> jepler: are you around?
[15:59:11] <awallin> lerman: i've been thinking about simulating milling...
[15:59:47] <lerman> Me too... for a long time. Do you mean showing the material that remains after cutting?
[15:59:57] <awallin> how hard is it to hijack the stream of canon-moves from the interp, not pass them to emcmot but somewhere else
[16:00:19] <lerman> Python seems to do that.
[16:00:26] <alex_joni> awallin: not _that_ hard
[16:00:29] <awallin> you mean AXIS
[16:00:37] <alex_joni> awallin: actually quite easy
[16:00:40] <lerman> Yes., of course.
[16:01:12] <alex_joni> awallin: look at canterp.cc
[16:01:14] <awallin> additionally, it would require some info about the stock
[16:01:40] <awallin> and then the hard part, a reasonably efficient and accurate stock model that's update when the cutter moves
[16:01:47] <lerman> Yes -- you are thinking about the same thing I've been thinking of.
[16:02:17] <lerman> One way is to use octrees to represent the 3-D model.
[16:02:48] <awallin> I'm not familiar with those, is it a kind of triangulation?
[16:03:04] <lerman> Are you limiting this to a 3 axis mill?
[16:03:34] <awallin> probably at first, but how much simpler does the problem get in 3D compared to 5D?
[16:03:40] <lerman> Octrees use a tree with 8 children at each level.
[16:03:46] <awallin> I mean cutter movement in 5D ;)
[16:04:29] <awallin> ok, I imagine 8^n will be a big number pretty fast...
[16:04:35] <lerman> A 3 axis mill lets you have s simple model -- 2-1/2 D. At each XY point, carry one value, the height.
[16:05:06] <awallin> lerman: yes, if the model is "pixel" based
[16:05:13] <lerman> Yes. So, octrees only subdivide when it is necessary -- at the boundary between to values (air or metal, in this case).
[16:05:24] <awallin> that sounds much better
[16:05:42] <lerman> You could use quadtrees in the 2-1/2 D case.
[16:05:42] <awallin> can the cutter shape also be represented with an octree?
[16:06:24] <lerneaen_hydra> even with 3-axis just a heightmap isn't enough, for example a disc cutter that cuts paralell to X/Y
[16:06:41] <lerman> A 10 inch by 10 inch stock, with .001 resolution is 100 million points.
[16:07:29] <awallin> lerman: but you just said you don't have to subdivide at some resolution. I would imagine a square piece of stock takes very little data to represent before any cut is made
[16:07:35] <lerman> Well. for 3 axis, I meant a "conventional" mill with a z-axis perpendicular to XY and the tool rotating around the Z axis.
[16:07:45] <awallin> lerneaen_hydra: yes, you mean a slot-mill, that would be a special case...
[16:08:11] <lerneaen_hydra> then again, that's quite rare and the simulator does need to (IMO) take every little possible thing into account
[16:08:18] <lerman> Ah. LH, I didn't catch what you meant. You are correct.
[16:08:52] <lerneaen_hydra> maybe defining toolshape as gcode would be easy to interpret?
[16:08:56] <lerman> Even threading with a thread mill would require more than just a 2-1/2 model.
[16:09:02] <lerneaen_hydra> ie a swept profile
[16:09:06] <awallin> lerman: have you looked at other volume representations, besides octrees?
[16:09:35] <maddash> has anyone here tried cradek's REALIZE script?
[16:09:44] <lerman> Well, the other natural one is CGS. But that can be expensive to implement.
[16:09:54] <awallin> maddash: what does it do?
[16:10:26] <maddash> awallin: autocad lisp script that exports gcode...can't get it working, though
[16:11:24] <awallin> maddash: oh, I haven't tried it
[16:11:59] <lerman> I once implemented a 2D model of a stroked brush. The shape of the brush would be specified, together with its path, and a drawing made in 2D of the results. Clever coding lets you just look at the pixels on the leading edge of the brush. That resulted in code that ran linearly in the diameter of the brush rather than as the square of the diameter.
[16:12:31] <awallin> lerman: I would think a good implementation needs to be adaptive, i.e. a square piece of stock requires very little storagem and as the tool cuts in to the material the model gets refined
[16:14:12] <lerman> That is what octrees would do.
[16:14:16] <lerneaen_hydra> hmm, might not this already be done and ripe for the picking from some OSS cam package?
[16:14:40] <awallin> lerneaen_hydra: and how many good oss cam packages do you know of? ;)
[16:14:52] <lerneaen_hydra> afaik there are ones that have a simulator
[16:15:04] <lerneaen_hydra> maybe easier than redoing it all. maybe ;)
[16:15:08] <awallin> lerman: ok, I will have to read about octrees
[16:15:09] <lerman> The model starts as a single node. When the first cut is taken, the node is broken into eight, and each of those nodes that are touch are also broken down, recursively.
[16:16:04] <lerman> BRLCAD is an open source package that does CSG. I suspect, though that CSG would be way too slow without a lot of special purpose code.
[16:16:19] <lerman> CSG => constructive solid geometry
[16:16:58] <lerman> At each location of the cylindrical mill, subtract the cylinder from the stock.
[16:18:26] <awallin> g0 g1 could probably done by subtracting the convex hull of the start and end points of the tool
[16:18:46] <lerneaen_hydra> lerman: if you do a slot, will the result be a rectangular slot or scalloped becuase of aliasing errors?
[16:19:01] <lerneaen_hydra> awallin: hmm, that sounds smart
[16:19:05] <lerman> I spent a little time looking at the data problem from a different point of view. Think of the problem as one of representing the surface of the stock. Then the problem is one of representing N squared points -- those on the surface, rather than representing the volume -- N cubed points.
[16:19:10] <lerneaen_hydra> relatively simple
[16:20:19] <lerman> Well, if you do the CSG approach, you can have a model without aliasing errors. You only see the errors when you output to a display, and that's controlled by the resolution of the display.
[16:20:35] <awallin> so, are there algorithms for calculating unions ad differences of polyhedra?
[16:20:50] <lerman> An octree model would have inherent errors based on the specified resolution.
[16:21:20] <lerman> Yes. There are such algorithms.
[16:22:04] <awallin> so, the surface of the stock could be triangulated, then for each move, we calculate a new triangulated stock surface
[16:22:21] <awallin> the tool would also be a triangulated surface
[16:22:27] <lerman> google constructive sold geometry.
[16:23:12] <lerman> although in this case, it would be destructive solid geometry :-) the algorithms generally include union and intersection
[16:23:53] <lerman> In its most general case, surfaces are NOT used. Solids are used.
[16:24:26] <lerman> If you use surfaces specified by triangles, you introduce aliasing at that level.
[16:25:27] <lerman> (Although there are cad systems that use surfaces.)
[16:26:04] <lerman> Bsplines, nurbs, etc, are commonly used.
[16:28:03] <lerman> See: http://www.melax.com/csg/index.html for a modeling program.
[16:28:59] <lerman> Remember, too, that in addition to a good modeling program, you need a good viewing program capable of displaying, rotating, scaling, etc the model that has been generatated.
[16:29:41] <SWPLinux> openGL should help with that
[16:30:14] <jepler> reminds me of this library (which I haven't actually used): http://www.opencsg.org/
[16:30:31] <awallin> yep, viewing things using openGL is not that hard...
[16:30:56] <lerman> jepler: been looking for you.
[16:31:00] <jepler> lerman: uh oh
[16:31:35] <jepler> lerman: I mean, what's up?
[16:32:40] <lerman> Tkinter questions:
[16:32:41] <lerman> 1 -- How do I implement the bar with the buttons that is just below the menu bar in axis. I notice the vertical separators?
[16:32:43] <lerman> 2 -- How do I make the menu bar not stretch when I strecth the window?
[16:34:30] <awallin> hmm, three 100W/3000rpm AC servos + drives, 550eur: http://riemun.huuto.net/auctionimages/a/ad/3abc37b4c6e93f412dced146ade80-orig.jpg
[16:34:45] <awallin> lerman: what are you programming? :)
[16:35:24] <lerman> GWiz -- a conversational gcode generator.
[16:36:11] <lerneaen_hydra> awallin: I thought you didn't want AC servos?
[16:36:19] <awallin> oh, I've seen it on some webpage
[16:36:55] <awallin> lerneaen_hydra: I'm all set for the mill, but you never know when those might come in handy! :) a lathe would be the next project
[16:37:05] <lerneaen_hydra> ooh, nice
[16:37:12] <lerneaen_hydra> want to trade? ;)
[16:37:52] <awallin> lerneaen_hydra: hmmm, I don't think the mill is for sale/trade
[16:38:04] <lerman> See: http://www.se-ltd.com/~lerman/gallery/G-Wiz -- the prototype is written in Qt, but I thought that python might be nice for this.
[16:38:05] <lerneaen_hydra> darn ;)
[16:39:55] <lerman> I need some servos and some linear slides for my lathe conversion. -- basically and XY table with a travel of around two feet on each axis. I already have some ball screws and bearings for them.
[16:40:05] <awallin> lerman: looks nice, it's a step towards CAM, but not quite freeform 3D yet...
[16:40:51] <lerman> That's correct. It handles common milling operations, though, in an easy to use manner.
[16:41:44] <lerneaen_hydra> hey, neat
[16:41:51] <lerneaen_hydra> nice for the common operations
[16:42:14] <lerneaen_hydra> lerman: OSS?
[16:43:12] <lerneaen_hydra> "...klermans lib..." looks like OSS
[16:43:30] <SWPLinux> written by klerman, for use with emc2 ... :)
[16:43:45] <SWPLinux> looks good to me
[16:43:46] <lerman> OSS? Not familiar with it.
[16:44:33] <jepler> lerman: if I recall correctly, the vertical separators are frames with "groove" relief
[16:44:42] <lerman> The neat part is that anyone can add to the library WITHOUT programming in anything other than gcode. Write a gcode subroutine.
[16:44:43] <jepler> lerman: (sorry, I keep getting called away from irc by my real job)
[16:44:50] <lerneaen_hydra> lerman: OSS = open source software?
[16:44:52] <lerneaen_hydra> err
[16:44:58] <lerneaen_hydra> s/?/.
[16:45:04] <lerman> Yes. Definitely open source.
[16:45:32] <lerneaen_hydra> lerman: oh, adding cycles in pure gcode sounds nice
[16:45:33] <jepler> lerman: to keep your toolbar from expanding vertically, use 'fill=X, expand=NO' when you pack it
[16:45:43] <lerneaen_hydra> what about adding the gui stuff?
[16:45:45] <lerman> jepler: working for a living is OK.
[16:46:13] <lerman> I'm not using pack. I'm using grid. Can I use both? Should I use pack?
[16:46:18] <lerneaen_hydra> lerman: it's not like there's much of a choice :p
[16:47:40] <jepler> * jepler re-reads what lerman said
[16:47:56] <jepler> lerman: is there something in the window that *should* expand when the window is resized vertically?
[16:48:18] <jepler> lerman: if so, find its row number, and in the containing widget do .rowconfigure(ROW_NUMBER, weight=1)
[16:48:32] <lerman> Probably the text area.
[16:48:55] <jepler> now the items in that row will grow, instead of the default (which seems to be to add empty space around the outside)
[16:49:05] <lerman> I tried setting the menubar weight=0, but that didn't seem to work.
[16:49:07] <jepler> the same should be done for .columnconfigure(COLUM_NUMBER, weight=1)
[16:50:23] <lerneaen_hydra> hi maddash
[16:50:31] <lerneaen_hydra> made any cakes? ;)
[16:50:53] <lerman> TclError: can't pack ".23475984": it's a top-level window
[16:51:48] <awallin> widget.pack() should work
[16:52:24] <maddash> hi lerneaen_hydra - I'm too blunt to unwrap any subtle entendres, so you're going to have to be as direct as possible
[16:53:02] <lerneaen_hydra> have you made any cakes with your cake decorating machine that goes faster than a speeding bullet
[16:53:13] <maddash> do you guys know what's wrong with the following? http://pastebin.ca/317420 when I pipe it through to mini, I get "unknown gcode used" -- ???
[16:53:37] <awallin> remove the % at the beginning?
[16:54:15] <maddash> awallin: don't think that's it, the error persists...
[16:54:17] <jepler> lerman: you can't pack() a Menu or a Toplevel or a Tk
[16:54:20] <jepler> (or grid them either)
[16:54:58] <lerneaen_hydra> brb
[16:55:15] <awallin> maddash: does it run in AXIS? that might give you a better error message
[16:55:33] <jepler> you make a menu into the menubar of a Toplevel or a Tk by doing app.configure(menu = ...)
[16:55:39] <SWPLinux> G00 F05 may be an error since no axis is specified for motion
[16:55:48] <jepler> SWPLinux: no, that should be accepted
[16:55:52] <SWPLinux> ok
[16:56:15] <jepler> It's G32 that is the problem
[16:56:18] <jepler> I don't believe emc defines G32
[16:56:26] <SWPLinux> those are awfully big numbers, too
[16:56:46] <jepler> http://linuxcnc.org/docs/2.1/html/gcode.html
[16:56:54] <SWPLinux> unless your UNITS are microns
[16:57:11] <maddash> hah.
[16:57:23] <maddash> hmm, so g32 must be vestigial
[16:57:58] <maddash> over here (http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?action=history&id=Gcode) it seems to be replaced by g33
[16:58:02] <CIA-8> 03jepler 07TRUNK * 10emc2/debian/ (emc2-docs.files rules.in): include gcode quick reference html in docs package
[16:58:33] <SWPLinux> hmmm - apparently it's a thread cutting code in some places
[16:59:08] <awallin> jepler: see how using the ctrl-" quotes fixed the missing end tag problem http://linuxcnc.org/docs/devel/html/hal/vcp/index.html
[16:59:36] <jepler> awallin: it looks better
[17:00:23] <jepler> maddash: m15 and m17 are also not defined by emc2
[17:01:36] <jepler> and emc2's F is clearly not the same F as in this program: at 5 units per minute it will take forever to go from 366204 to 475527 units
[17:01:41] <maddash> jepler: that's odd. I used this (http://www.mgl.ca/~ecp/gobweb/cnc.c) linked to by (http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Cam) on a simple dxf with 1 square and 2 circles
[17:01:49] <jepler> so basically the problem with that g-code is that it's a completely incomptible dialect
[17:02:00] <cradek> this must be for a very old machine with implied decimal points
[17:02:44] <maddash> cradek: no, those circles really are that big
[17:04:15] <maddash> wow, I'll actually have to write my own code. this is bad.
[17:04:29] <jepler> maddash: I added a note to the wiki that that program doesn't output emc compatible g-code
[17:05:06] <cradek> what are you trying to do?
[17:05:44] <maddash> cradek: simple dxf to gcode...
[17:06:19] <cradek> do you have autocad?
[17:06:25] <skunkworks> Our really old k&t didn't use decimal places - relied on leading zeros for place holders
[17:07:07] <maddash> I can't get your REALIZE to work...it just outputs a 57-byte C:\acad.t (simple preamble and footer) that doesn't do anything...
[17:07:33] <cradek> did you read all the instructions? did you do the example program on the web page?
[17:08:20] <maddash> example program? you mean the rectang done inside 250-MILL?
[17:08:25] <cradek> yes
[17:08:54] <maddash> did that one, and variations of it (125-MILL, vfeed: 11, polylines, etc) already
[17:09:12] <CIA-8> 03jepler 07v2_1_branch * 10emc2/debian/ (emc2-docs.files rules.in): merge from HEAD: include gcode quick reference html in docs package
[17:09:52] <maddash> I'm pretty sure I missed something, but it wouldn't be anything on the webpage, or any of the comments in the lisp code...
[17:10:08] <cradek> are you sure your rectang is at z -.1 and on layer 250-MILL?
[17:10:25] <cradek> if it's at z 0 or on layer 0 it is ignored
[17:11:29] <cradek> LIST your rectangle and see what the points and layer are
[17:12:02] <jepler> lerman: http://pastebin.ca/317446
[17:12:19] <jepler> lerman: the Vrule class and the arguments to .pack give a vertical bar like the separator on the axis toolbar
[17:12:25] <lerneaen_hydra> iab
[17:12:36] <lerman> Thanks.
[17:13:03] <jepler> it should look something like this: http://emergent.unpy.net/files/sandbox/separator.png
[17:13:37] <maddash> cradek: ok, I created a new file, then "RECTANG"-ed with (0,0,-1) - (1,1,-1) inside layer "250-MILL"
[17:13:58] <maddash> cradek it should be ok to run realize now, right?
[17:14:16] <cradek> yes
[17:14:23] <lerneaen_hydra> maddash: I hope your units are mm, and not inch
[17:14:36] <cradek> doesn't matter for a test
[17:14:58] <maddash> cradek: can I stay with the default options (0.02, etc)?
[17:15:04] <cradek> yes
[17:15:18] <cradek> and when it asks for a selection, select the rectangle (or type "all")
[17:16:25] <cradek> ideally it should generate .ngc by default now, I'll change that
[17:16:53] <maddash> yes, did all that, and I still get the do-nothing file - "[...]G00 z1\n M02"
[17:17:37] <jepler> lerman: to get a button that is flat until moused over, use relief=Tkinter.FLAT, overrelief=Tkinter.Raised
[17:17:44] <jepler> er, Tkinter.RAISED
[17:17:46] <cradek> what autocad?
[17:18:20] <maddash> 2002
[17:18:34] <cradek> does the rectangle explode if you try it?
[17:18:43] <maddash> do I have to add the TEXT and "vfeed: 10" first?
[17:18:50] <cradek> no, it will use a default
[17:18:53] <maddash> no, no explosion
[17:19:09] <cradek> does it give an error when you run realize?
[17:20:09] <cradek> are you selecting the rectangle, or typing "all"?
[17:20:49] <maddash> none, a `tail` of the cmd line gives: (http://pastebin.ca/317456)
[17:21:16] <maddash> I select the rectangle, but that's equiv to all, b/c in both cases only 1object is returned
[17:21:35] <cradek> oh it's a LWPOLYLINE
[17:21:49] <cradek> see the first "helpful hint" on the webpage
[17:24:03] <maddash> aha. so I'll have to do this everytime I want to REALIZE?
[17:24:22] <alex_joni> depends what you want to realize :)
[17:24:29] <cradek> I think once per drawing, not sure
[17:27:11] <skunkworks> maddash: are you running a microsoft machine? http://www.dakeng.com/ace.html -- It has been a while but I remember having decent luck with it also. Have not tried REALIZE although I want to.
[17:27:50] <skunkworks> looks like it has been updated since I have used it.
[17:28:54] <stan> hey all
[17:28:55] <maddash> skunkworks: the ace program executes a segmentation fault everytime I feed it a dxf...
[17:29:20] <skunkworks> ouch
[17:29:28] <jepler> http://blog.modernmechanix.com/2007/01/15/ad-rca-color-tv-with-solid-copper-circuit-dependability/
[17:29:49] <jepler> (they mean printed circuit boards)
[17:29:49] <alex_joni> maddash: never heard about segmentation fault on doze before
[17:30:22] <lerneaen_hydra> alex_joni: afaik there are "posix" extensions for windows
[17:30:41] <maddash> alex_joni: sry, meant the usual "access violation"
[17:30:51] <cradek> jepler: dependable right until the heat under the tube sockets turns the PCB into dust
[17:31:04] <alex_joni> lerneaen_hydra: yes, there were.. but I don't think those are used/present anymore
[17:31:16] <lerneaen_hydra> afaik they are actually available in windows
[17:31:18] <lerneaen_hydra> xp
[17:31:19] <alex_joni> lerneaen_hydra: at one point there was a *nix by microsoft
[17:31:25] <lerneaen_hydra> if you install unix services or something
[17:31:28] <alex_joni> integrated in NT or so
[17:31:38] <lerneaen_hydra> yeah, they ported that to 2k, and then xp
[17:32:17] <alex_joni> xenix?
[17:32:21] <alex_joni> something like that
[17:32:57] <stan> can anyone point me to some documentation on controlling the spindle speed/rotation with emc2?
[17:33:35] <SWPLinux> xenix was from the '70s, not in NT
[17:34:00] <SWPLinux> MS has "UNIX services for Windows NT", which is probably still available (under a slightly changed name)
[17:34:37] <jepler> stan: this page describes the HAL pins which give the spindle speed and direction: http://linuxcnc.org/docs/2.1/html/config/emc2hal/index.html
[17:34:53] <jepler> some of it may apply only to the upcoming emc2.1, but most should be valid for emc 2.0.x.
[17:35:15] <lerneaen_hydra> * lerneaen_hydra what SWPLinux said
[17:36:38] <jepler> stan: motion.spindle-on motion.spindle-speed-out etc
[17:37:00] <stan> jepler: thanks.
[17:37:59] <alex_joni> http://en.wikipedia.org/wiki/Image:Ms_xenix.png
[17:38:08] <stan> I was hoping for more outputs. :-) Like maybe some PWM output or something...
[17:38:20] <alex_joni> stan: you can link the speed-out to anything you want
[17:38:28] <alex_joni> for example a free DAC on your board
[17:38:42] <SWPLinux> stan: you can generate PWM, steps, analog (with suitable hardware), quadrature ...
[17:38:42] <alex_joni> or to the input to a freqgen component
[17:39:07] <alex_joni> you can set up freqgen to get the input from the spindle-speed-out, and generate a PWM which you link to a parport pin
[17:39:24] <lerman> jepler: it appears that grid and pack don't play well together. (seems to hang using 100% of the cpu). Have you tried both? Is pack preferred to grid? Any particular reason?
[17:39:26] <stan> perfect! That's exactly what I need.
[17:39:32] <jepler> stan: the "nist lathe" turns the spindle-speed-out signal into a pseudo-PWM signal, which may be what you want: http://cvs.linuxcnc.org/cvs/emc2/configs/nist-lathe/
[17:40:05] <jepler> lerman: yes that's a long-standing bug in Tk that they refuse to fix--you can get lockups if you grid and pack inside the same frame
[17:40:38] <jepler> lerman: I use pack when there is a simple left-to-right or top-to-bottom layout, and grid otherwise.
[17:41:26] <lerman> jepler: also, you import Tkinter. I've been from Tkinter import *. Your usage requires Tkinter.Button.. -- the alternative is just Button... Is one form preferred over the other? (Why?)
[17:41:36] <Dallur> maddash: is that dxf file something you can share? I'm building a dxf to gcode program and I would like to test your file (especially if it crashes other programs)
[17:41:49] <lerman> I should get this right before I write too much code.
[17:42:12] <awallin> lerman: you clutter up the namespace with from module import *
[17:42:22] <jepler> lerman: if you are writing a module that will be 'import'ed, then you should avoid 'import *' because it "pollutes the namespace"
[17:42:29] <jepler> lerman: if you're writing a main program then go ahead and 'from Tkinter import *'
[17:43:21] <lerman> Thanks. Works for me. I'll probably change to not pollute the namespace -- even though this is a main program (today).
[17:49:36] <awallin> lerman: would you be interested in a cnc simulation paper I found?
[17:50:18] <lerman> awallin: Sure...
[17:50:33] <maddash> Dallur: http://pastebin.ca/317501
[17:51:08] <Dallur> maddash: thx, I will try it
[17:51:34] <maddash> guys, is it just me, or is there no source to this program (http://cnccodegen.sourceforge.net/) linked to by the "Cam" page from wiki.linuxcnc.org ?
[17:51:55] <awallin> lerman: can you recieve the pdf by dcc, or should I email it to you?
[17:52:36] <lerman> I'm loading it now.... Got it. Thanks.
[17:52:46] <maddash> jepler: http://jelinux.pico-systems.com/gcode.html seems to be dead. should it be removed from the wiki?
[17:53:06] <awallin> lerman: oh? in my end it looks like the transfer hasn't even started...
[17:53:27] <alex_joni> maddash: no, the pico-systems is down at the moment
[17:53:33] <alex_joni> but it's very much working usually
[17:53:41] <lerman> Well, it looks like I got it, but I can't open it.
[17:53:44] <maddash> ok.
[17:54:01] <awallin> size 0b ?
[17:54:02] <jepler> maddash: I don't see any source to "cnccodegen" either, but I didn't look very hard
[17:54:30] <SWPLinux> on the right hand side is a "download" link
[17:54:40] <lerman> email to: lerman-awallin at se-ltd.com (yup zero size).
[17:54:45] <SWPLinux> it's a zip file, which contains some java code, among other things
[17:55:46] <awallin> lerman: ok, sent it
[18:00:51] <lerman> awallin: got it.
[18:01:38] <awallin> lerman: a 2D simulation might be simpler to start with. it might have applications in generating pocketing toolpaths for a CAM package too
[18:06:04] <alex_joni> maddash: didn't find any source either, only java classes.. but those can be easily decompiled
[18:13:19] <lerneaen_hydra> awallin: it's not very difficult to go from heightmap to 3d
[18:13:35] <lerneaen_hydra> even with aliasing issues
[18:24:53] <lerman> awallin: started reading the paper. The GWB representation seems interesting.... But the whole thing seems too complicated to do overnight.
[18:25:41] <lerman> My view is that if you can't get something running in a few days, it won't get done (at least not by one person).
[18:28:11] <awallin> lerman: you are probably right, one weekend of work must result in a 'wow' effect for a number of other developers...
[18:31:35] <lerman> So, if we divided the 8 x 10 x 10 volume into voxels of .01 inch, we would have 8 x 10^8 voxels. With 1 bit per voxel, that would be 10^8 bytes. 100 megabytes fits into ram quite nicely these days. A resolution of .01 inches isn't great, but would certainly give a nice proof of concept.
[18:32:08] <fenn> awallin: http://verot.cvs.sourceforge.net/verot/freesteel/?hideattic=0
[18:32:35] <awallin> hmm, 0.01 inch = 0,254 mm
[18:33:04] <lerneaen_hydra> IMO that resolution is quite good if you take the total dimensions into account
[18:33:05] <fenn> no idea why they deleted most/all of the files, but there they are
[18:33:18] <awallin> fenn: I've seen that, but there are no docs as far as I can tell, so it's a bit hard to just jump in the code
[18:33:33] <fenn> yeah it was definitely a work in progress
[18:33:45] <awallin> it seems to be done by the freesteel guys? but abandoned 2 years ago
[18:34:22] <awallin> lerman: with a voxel data-structure, you would still have to extract the surface somehow for graphical display
[18:34:38] <fenn> * fenn mumbles something about octrees
[18:34:54] <lerman> I'm sure there is a good zbuffer algoritihm to use.
[18:35:04] <lerman> fenn: that's where I started.
[18:35:29] <fenn> i dont actually know what y'all are doing
[18:35:47] <fenn> 2.5d cam?
[18:35:51] <lerman> The problem is that these techniques grow as the size of the volume -- but we only see the surface. So, a method that grows as the size of the surface is better.
[18:36:18] <lerman> Simulation of cutting -- show the material remaining -- taking into account the form of the tool.
[18:36:29] <awallin> fenn: cnc simulation, i.e. from g-code, show the stock/part as it is being cut
[18:36:37] <lerman> Start with a 3 axis mill.
[18:37:06] <lerman> Axis shows the cutter path. Give the cutter a shape and have it remove material.
[18:37:21] <SWPLinux> you don't necessarily have to represent the entire envelope of the machine
[18:37:33] <SWPLinux> 8x10x10 is a pretty big hunk of steel
[18:37:34] <lerman> Watch it cut in real time. Be able to zoom/translate/rotate the material.
[18:37:43] <lerneaen_hydra> being able to model the vise would be nice
[18:37:48] <lerman> Sure. I just picked some numbers.
[18:37:55] <lerneaen_hydra> cross sections would be nice too
[18:37:59] <lerman> That would fit in memory.
[18:38:10] <SWPLinux> my memory isn't that big ;)
[18:38:18] <fenn> didnt twingy have something that already works?
[18:38:19] <lerman> Yes -- zoom/translate/rotate/slice.
[18:38:22] <lerneaen_hydra> the vise doesn't need to be volumetric though
[18:38:40] <SWPLinux> I wonder if parasolid or something similar is available in open source
[18:39:14] <lerman> The vise resolution isn't very important -- .1 inch (.25mm) would be fine.
[18:40:15] <lerneaen_hydra> it seems to be a waste to do volumetrics with the vise, it should just turn red or something if the mill touches it
[18:40:28] <lerneaen_hydra> just a surface
[18:40:29] <lerman> I could do a crude implementation using voxels in a few days. I doubt you^H^H^H^H I could do one with parasolids or whatever in that time.
[18:40:42] <SWPLinux> I agree with that doubt
[18:41:01] <lerneaen_hydra> that's true
[18:41:04] <lerman> But do you agree with a few days using voxels?
[18:41:19] <fenn> zbuffer is probably more promising
[18:41:27] <SWPLinux> for a simplistic approach, maybe
[18:41:53] <fenn> voxels dont scale well at all
[18:42:09] <awallin> voxels would probably be doable yes, finding the surface and plotting it using opengl might require a few more days
[18:42:27] <SWPLinux> at this point, the concern LH voiced (regarding slitting cutters) can't be represented in emc, so the height map / octree approach probably isn't bad
[18:42:43] <lerman> Well, first get a proof of concept -- make it work. Then make it work well. Then make it work fast. Then convert to octrees (because they use voxels in a way that runs proportionally to surface area).
[18:43:03] <fenn> i bet there's a lib out there for rendering voxel data (marching cubes algorithm or similar)
[18:43:05] <lerneaen_hydra> oh, octrees is voxel based?
[18:43:09] <lerman> I seem to remember something called the "marching cube algorithm" that addresses display issues.
[18:43:11] <fenn> lerman: no
[18:43:17] <fenn> argh
[18:43:20] <lerman> shit fenn, you type faster than me.
[18:43:23] <fenn> lerneaen_hydra: no
[18:43:28] <fenn> * fenn kicks his tab key
[18:43:34] <SWPLinux> in fact, I don't think emc has the ability to have the tool shape in the tool table (flat, ball, bull-nose, V ...)
[18:43:41] <SWPLinux> (except for lathes)
[18:43:41] <fenn> it should just beep at you if there's multiple completions
[18:43:54] <lerman> Well, octrees are voxel based, but the voxel size varies.
[18:43:56] <lerneaen_hydra> SWPLinux: that could be added ;)
[18:44:02] <lerneaen_hydra> lerman: ooh, nice
[18:44:04] <awallin> SWP: probalby only length and diameter right now
[18:44:11] <alex_joni> SWPLinux: patches gladly accepted :)
[18:44:17] <lerneaen_hydra> lerman: they vary in realtime?
[18:44:20] <lerman> Easy to add more stuff to the tool table.
[18:44:21] <lerneaen_hydra> (UI realtime)
[18:44:22] <SWPLinux> it could be, but making something that allows for slitting saws is - er - complex :)
[18:44:42] <lerneaen_hydra> IMO it would be easy to define the toolshape with a kind of gcode
[18:44:54] <SWPLinux> easy, but not necessarily useful ;)
[18:44:56] <lerneaen_hydra> the swept profile of the cutter
[18:45:10] <alex_joni> a 3D model is always better
[18:45:10] <fenn> there's only so many different cutter shapes
[18:45:11] <lerneaen_hydra> or maybe an image with a B/W mask
[18:45:22] <fenn> i think a conical cutter encompasses almost all possible endmill parameters
[18:45:26] <SWPLinux> fenn: tell that to guys using router bits
[18:46:25] <lerman> Yeah -- sort of. An octree starts as a single cube. As soon as it is no longer uniform, it is broken into eight (hence "oct") subtrees. Some of those are uniform. Others will be subdivided as they are hit by the cutter. Uniform in this context means of a constant material (air or metal). You keep subdividing down to the target resolution.
[18:46:59] <SWPLinux> it's basically voxels with 3D RLL compression
[18:47:14] <SWPLinux> err - RL, not RLL
[18:47:24] <lerman> For now, I would just add another table that is outside of emc to relate a cutter number (inside emc) to a cutter form model.
[18:47:28] <lerneaen_hydra> lerman: when they subdivide, do they do that into cubes of equal size?
[18:47:37] <SWPLinux> lerneaen_hydra: no
[18:47:45] <lerneaen_hydra> ok, good
[18:47:45] <SWPLinux> (I don't see how that could work, so I say no :) )
[18:47:54] <lerneaen_hydra> err
[18:47:55] <awallin> lerneaen_hydra: http://hpcc.engin.umich.edu/CFD/users/charlton/Thesis/html/node29.html
[18:47:59] <lerman> Yes, the subdivisions are equal size.
[18:48:27] <SWPLinux> the reason is that if you cut off a 0.00001 sized chink, you'd need to subdivide the whole thing into 0.00001x0.00001x0.00001 chunks to represent that air
[18:48:33] <lerneaen_hydra> hmm, seems wasteful, can cubes be rejoined to make rectangles?
[18:48:39] <lerman> At least in the version I'm familiar with. BTW quadtrees have been around longer. They are used on the plane.
[18:50:04] <fenn> * fenn would like to request arbitrary 3d stock models while we're at it
[18:50:09] <SWPLinux> eek. that means you get a lot of work the first time a tool skims the part
[18:50:22] <lerman> They can be rejoined to make the original cubes. So once an entire cube is cut out, its components are now of the same material (air) and can be rejoined. To be efficient, that should be done. The only broken cubes should be on the cut surface.
[18:50:41] <lerman> SWPLinux: that is true.
[18:50:44] <lerneaen_hydra> lerman: yeah, that's what I was thinking
[18:51:03] <SWPLinux> lots of recursive work to get the surfaces right
[18:51:07] <lerneaen_hydra> it seems that ram use will fluctuate wildly when doing all the dividing
[18:51:21] <lerneaen_hydra> and if you run out of ran... nastyness (swap)
[18:51:43] <lerman> To repeat myself (somewhat differently) -- the key to making this work is to have storage requirements that grows as the cut surface area -- NOT as the material volume.
[18:52:03] <SWPLinux> octrees seem to be a hybird there though
[18:52:06] <SWPLinux> hybrid
[18:52:18] <lerman> It doesn't have to fluctuate. Just allocate all you have when you start.
[18:52:38] <SWPLinux> heh - 100% memory usage for the tiniest parts ;)
[18:53:12] <fenn> doesnt the kernel handle all that stuff?
[18:53:14] <awallin> lerman: but octrees would be volume representations, right? so the storage requirement would grow as the volume with an octree
[18:53:33] <SWPLinux> only the volume that contains a metal/air surface is represented
[18:53:57] <lerman> awallin: yes. But large cubes would be used internally. Only the part that are on the interface are subdivided into small parts.
[18:53:59] <SWPLinux> plus the rest of that octant (at whatever resolution you represent the cut)
[18:54:02] <fenn> cubical shapes would take less memory than, say, spherical shapes
[18:54:20] <lerman> fenn: true -- but not enough to really matter.
[18:54:28] <lerneaen_hydra> if you start with a cube and then make a slot memory use would first increase, and then decrease when the cubes are reconnected
[18:54:29] <fenn> no?
[18:54:46] <lerman> lernean... correct.
[18:56:23] <SWPLinux> a hemisphere should be in the same order of magnitude as a simple voxel representation (off by some factor, but the same O)
[18:56:29] <lerman> fenn: cubical shapes only save memory if the edges are on the boundaries of large cubes. But the probability of that is relatively small. I'll bet you can prove that the difference is only a factor of two or so. That's not a lot by my measures.
[18:57:15] <lerman> Yes SWP. The big 'O' is what matters.
[18:57:26] <SWPLinux> octrees are O(surface complexity^3), vs. voxels being O(dimensions^3)
[18:57:26] <fenn> can't you scale the octree resolution to match the dimensions of the part?
[18:58:00] <SWPLinux> so an octree devolves into voxels for pathological cases
[18:58:14] <SWPLinux> (plus overhead, but that's a constant factor)
[18:58:28] <lerman> I know what dimensions are (10 x 10 x 10) but the measure of surface complexity isn't clear.
[18:58:56] <SWPLinux> right, and as you pointed out, if the box edge deosn't fall on a voxel box edge, then you have a problem
[18:59:09] <lerman> I would say Octrees -- O(surface area). voxels -- O(volume).
[18:59:16] <SWPLinux> consider an 8x8x8 volume (for simplicity)
[18:59:33] <lerneaen_hydra> SWPLinux: the size of the boxes will decrease untill they do
[18:59:36] <SWPLinux> if you have a cube of metal 4x4x4, starting at 0,0,0, then you have a single division
[19:00:39] <SWPLinux> if that cube is 4.00001 per side, then you have a progressively smaller set of divisions, in the 7 "empty" spaces - the only one that remains pure is the one that's all metal
[19:00:53] <lerman> Well the probability of an edge (measured in .001s) falling on a .001 is 1. falling on a .002 (the next size) is 1/2 falling on a .004 is 1/4.
[19:01:13] <SWPLinux> and the sum of those approaches 1
[19:01:18] <lerman> And remember, you need to have a hit in all three dimensions.
[19:01:47] <SWPLinux> I'm trying to think through the requirements for a "slightly too big" block
[19:02:02] <SWPLinux> so in my example, 4.001x4x4 (to make it easier)
[19:02:12] <lerman> Yes. I know. That's where I came up with my factor of two. But that's wrong; the cube must hit those values on all edges.
[19:02:28] <awallin> SWP: but not all of the 7 empty spaces would have to divided into 0.001 voxels, just the boxes close to your 4.001 cube
[19:02:36] <lerman> I'm sure someone could derive the average values.
[19:02:51] <SWPLinux> right, which touches all 7 other octants (in the 4.001^3 case)
[19:02:52] <lerman> (But not me.) :-)
[19:02:55] <SWPLinux> heh
[19:03:06] <SWPLinux> I'm trying to use my IRC calculator to do it ;)
[19:03:12] <SWPLinux> (it's not workign)
[19:03:59] <lerman> Maybe when I was in college -- say 40 years ago. (When were octrees invented?) Hell, when were trees invented? The green leafy kind :-)
[19:04:33] <lerman> But you dont need the answer. Faith is the answer (what was the question again)?
[19:04:54] <SWPLinux> ok - for a flat face x by x, you have sum(x ^2 + (x/2)^2 + (x/4)^2 ... )
[19:05:15] <SWPLinux> so it is actually a factor of 2, for a flat surface :)
[19:05:27] <lerman> I'm sure octrees would work. I'm not sure how hard they would be to use for this, though. Also, they may be inefficient.
[19:05:32] <SWPLinux> err - nope. a bit less, but anyway
[19:06:06] <alex_joni> brb, sleep
[19:06:12] <SWPLinux> hasta
[19:06:17] <lerman> In terms of time; not space.
[19:06:21] <SWPLinux> yeah
[19:06:28] <SWPLinux> the problem is the recursive algorithm
[19:06:54] <awallin> bbl
[19:06:55] <SWPLinux> for 0.001 resolution, you may need 13 or 14 levels of recrsion to get it
[19:07:23] <SWPLinux> 10 inches needs 14-ish bits
[19:07:44] <SWPLinux> so you have 14 subdivisions
[19:07:48] <lerman> That isn't the real issue, I think. The real issue is that when you take small cuts, you break things apart and then have to reassemble them. (I think).
[19:08:17] <SWPLinux> jus tbreaking things takes time. putting them back together is another nightmare^W issue
[19:09:02] <lerneaen_hydra> SWPLinux: there are many different ways to put them together too
[19:09:03] <lerman> Well, memory is cheap. And virtual memory is even cheaper. Back to the voxel approach.
[19:09:03] <SWPLinux> hmmm. both may be doable in the same function
[19:09:08] <lerneaen_hydra> say for example you cut an arc
[19:09:24] <lerneaen_hydra> which cubes do you bunch together with each other?
[19:09:42] <lerneaen_hydra> maybe just subdivide and no recombine
[19:09:53] <SWPLinux> putting them together is possibly as simple as the visit() function returning a material type, and if they're all the same, delete the children and make the parent have the common type
[19:10:00] <lerman> You only break things in half on each axis. Then you put them back together the same way.
[19:10:10] <lerman> Exactly (SWP).
[19:10:29] <lerneaen_hydra> then again, if you think about the resources we've got. this is going to be running on RT stuff, so it seems to me that things are more CPU bottlenecked than RAM bottlenecked
[19:10:36] <Dallur> maddash: if you read logs, I tried your dxf and the reason it is failing with many programs is that the size of your objects is huge 9.232.004.100.369 for example, in mm that would be about 1500 times around the earth, in inches that would be just a bit farther away than the sun :P
[19:10:41] <lerneaen_hydra> probably best to go for filling the ram out
[19:11:06] <SWPLinux> lerneaen_hydra: I don't know that that's true
[19:11:17] <SWPLinux> the CPU issues aren't about throughput, they're about latency
[19:11:25] <SWPLinux> this is a throughput issue
[19:11:37] <lerneaen_hydra> yes but think about the cpu power left after all the RT bits
[19:11:44] <lerneaen_hydra> not that much (at least for me)
[19:11:47] <SWPLinux> that depends on the system :)
[19:12:18] <SWPLinux> on my celeron 500, things are fine, though a little slow, when doing software step generation (16 or 20 uS BASE_PERIOD)
[19:12:38] <lerman> Think of this as a stepper system. Each step (of say .001), you look at the edge of the tool and see what it is newly impinging on. Then you process those voxels. (in octree or other notation)
[19:12:39] <SWPLinux> when I make it use the Mesa / USC card, that'll be different
[19:13:14] <SWPLinux> the problem is that you'd ideally do that with the IMPUT_SCALE resolution (which may not be the same for all axes)
[19:13:29] <SWPLinux> though I guess it doesn't matter for visualization purposes
[19:13:31] <lerman> The real utility of this is not in real time mode (just look at what you are cutting, dummy). This is most useful for simulation before you hit the metal.
[19:14:07] <fenn> lerman: another "pure" approach is called swept volume, better than just plain csg because there is no time-slice aliasing
[19:14:15] <fenn> but i have absolutely no idea how to do that
[19:14:36] <SWPLinux> hmmm. in that case, you can just do all the work needed to generate GL models for the part, then render it using openGL ...
[19:14:49] <fenn> may be better for comparing the gcode output to the desired result, than for an animated stock simulation
[19:15:22] <lerman> I would animate and step thru the gcode. That lets you see where you caused the problem.
[19:15:48] <lerman> Of course, I would want a better step capability, too.
[19:16:00] <SWPLinux> http://gamma.cs.unc.edu/SV/
[19:16:27] <fenn> i was hoping that at some point cutter physics could be modeled and you could see where there's inaccuracies in the part due to a heavy cut
[19:17:12] <fenn> that would work with any other approach too
[19:17:36] <lerman> The paper awallin provided does some of that.
[19:17:46] <fenn> got a link?
[19:18:28] <lerman> Nope. He emailed it to me.
[19:20:21] <fenn> wish these academic papers would just publish the code
[19:20:31] <lerman> Don't we both.
[19:20:46] <SWPLinux> that would rob their universities of their rightful patent portfolio income
[19:20:47] <lerman> Even then, it might be a bitch to use.
[19:20:48] <fenn> "look what we can do! nyaah"
[19:21:17] <fenn> yes their rightful federal tax funded research profits
[19:21:50] <fenn> at least this one has a paper available for anyone to read online
[19:22:12] <lerman> I guess that we CAN talk. We open source a lot of stuff.
[19:22:15] <SWPLinux> heh - gotta love the "greed before need" mentality, huh? :)
[19:22:32] <lerman> ....but not everything.
[19:26:38] <SWPLinux> ooooh - it looks like the snow stopped. I now have the opportunity to rev up the snowblower
[19:26:40] <SWPLinux> bbl
[19:28:18] <skunkworks> we got about 6 inches here
[19:29:43] <SWPLinux> I wonder if someone should contact these people: http://web.umr.edu/~vrpl/proj-cam.htm
[19:32:49] <fenn> i sent a mail to them a year ago.. no reply
[19:33:15] <SWPLinux> http://www.sdpsnet.org/journals/vol4-1/cutter.pdf
[19:33:21] <fenn> i asked that they reply even if they didnt want to help me so i knew that they actually got my email..
[19:34:25] <fenn> such a waste
[19:40:27] <Dallur> maddash: I tried your dxf and the reason it is failing with many programs is that the size of your objects is huge 9.232.004.100.369 for example, in mm that would be about 1500 times around the earth, in inches that would be just a bit farther away than the sun :P
[19:42:37] <maddash> huh?
[19:43:06] <maddash> you mean, 249.9232004100369?
[19:43:17] <maddash> how's that 1500 times around the earth?
[19:43:39] <Dallur> maddash: sorry, I did a search/replace on the file because the copy from pastebin had line numbers :P
[19:43:59] <Dallur> maddash: that's still one hell of a resolution though
[19:45:33] <maddash> should be truncated and properly rounded, then
[19:45:59] <Dallur> maddash: probably not, I would be very mad if a dxf to gcode converter truncated my model
[19:46:19] <maddash> brb
[19:48:51] <daniel_br> Lerman: opensource cam http://www.comaxim.com/
[19:49:31] <lerman> thanks, I'm looking now.
[19:55:03] <lerman> Doesn't seem to be much there.
[19:56:26] <daniel_br> Yes, maybe you can e-mail him
[19:58:11] <daniel_br> and you can use some code from http://www.opencascade.org/ ? or not?
[20:02:27] <Dallur> maddash: is this what you wanted in g-code ? -> http://pastebin.ca/317675
[20:03:24] <fenn> opencascade looked really hard to use
[20:03:40] <fenn> but maybe that's just me
[20:04:07] <fenn> (not to mention it isnt actually free)
[20:05:00] <awallin> the verot project on sourceforge seems to have used vtk, and that's what the freesteel guys advocate in their blog also
[20:05:16] <awallin> http://www.vtk.org/
[20:06:28] <maddash> Dallur: what's that generated from?
[20:06:39] <Dallur> maddash: my own python script
[20:06:55] <awallin> vtk isn't under GPL either, but the source is available
[20:06:57] <Dallur> maddash: but there is quite a bit of stuff which is ignored, soft pointers and such which I have not implemented in my script yet
[20:07:19] <fenn> Modified source versions must be plainly marked as such, and must not be misrepresented as being the original software.
[20:07:20] <maddash> Dallur: is that the corresponding g-code to my dxf?
[20:08:14] <awallin> fenn: that's not so bad, is it?
[20:08:14] <fenn> well, that sounds reasonable
[20:08:15] <Dallur> maddash: yes and no, I ignore some stuff in your file because I have not implemented it in my script
[20:08:31] <maddash> Dallur: ah. what if I wanted to change the starting point?
[20:09:10] <Dallur> maddash: you could subtract the same number from all X values or all Y values
[20:10:50] <Dallur> brb, grabbing a bite
[20:13:54] <skunkworks> maddash: I just ran some dxf files through ace - seemed to work for me.
[20:14:14] <maddash> that's odd. wine?
[20:14:31] <skunkworks> no - xp
[20:14:44] <skunkworks> * skunkworks isn't really a linux person.
[20:14:50] <skunkworks> yet
[20:16:19] <maddash> yeah, powerstation's getting kind of old, but I haven't found a gpl equivalent yet
[20:18:26] <jepler> http://www.frozeneskimo.com/fpgaboards.html -- darn, the DS2B sounded good, but the actual price is $99, not $54 as shown in his list
[20:18:34] <jepler> not that I need another FPGA board, after pluto and mesa
[20:18:51] <fenn> buy more. buy more and be happy.
[20:19:57] <SWPLinux> jepler: http://www.picocomputing.com/products/cards.php
[20:20:23] <fenn> wow
[20:20:36] <daniel_br> one good priced amateur windoze cad/cam http://www.iprocam.com/ninoscadcam/index.html
[20:21:02] <daniel_br> fourth axis also
[20:21:27] <SWPLinux> I have CadMax Solid Master: http://www.cadmax.com
[20:21:55] <SWPLinux> it's actually a parametric solid modeller (similar in function, but not form, to SolidWorks)
[20:22:48] <SWPLinux> though it seems there haven't been any updates in oh - say 5 years
[20:22:48] <fenn> hmmm @ pastel colors
[20:24:45] <jepler> SWPLinux: interesting -- did you spot any prices?
[20:24:59] <SWPLinux> not on the E12/E14
[20:25:00] <ds3> for tiny 1liner patches, where should they be sent to?
[20:25:27] <fenn> here or developer's mailing list
[20:26:22] <SWPLinux> jepler: there was an article in the latest Linux Journal about them
[20:26:53] <ds3> src/Makefile: around line 610: should use @$(CC) instead of directly calling @gcc since that'd defeat the point of having configure
[20:27:46] <jepler> ds3: the rule for '../rtlib/%.so'?
[20:27:53] <ds3> yeah
[20:28:08] <jepler> I'll fix that
[20:28:40] <ds3> this is from me doing a 32bit compile on a 64bit box; I do it with CC="gcc -m32" ./configure and that line breaks this (FYI)
[20:31:12] <CIA-8> 03jepler 07v2_1_branch * 10emc2/src/Makefile: merge rev 1.192: use $(CC) instead of gcc
[20:31:29] <jepler> ds3: thanks, fixed now
[20:39:33] <daniel_br> another approach for g code generation http://www.rainnea.com/cnc_toolkit.htm
[20:39:57] <SWPLinux> that one is cool, but requires GMax to work (or possibly 3dsMax)
[20:40:24] <SWPLinux> it's more meant for artistic g-code (such as jewelry), rather than technical (such as CAD)
[20:40:38] <SWPLinux> it does do full 5-axis output though
[20:40:50] <ds3> always rhino3d + madcam
[20:41:04] <SWPLinux> I can't stand Rhino, personally
[20:41:23] <SWPLinux> I change my mind too often to use anything but a fully parametric modeler :)
[20:41:39] <ds3> mind giving your opinion? I been playing with the demo
[20:41:50] <SWPLinux> heh
[20:41:51] <ds3> how does being parametric help there?
[20:42:07] <SWPLinux> well, one of the Rhino demos is making a cone or something, right?
[20:42:10] <fenn> daniel_br: feel free to add anything that is missing to this page: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Cam
[20:42:12] <Dallur> I love Rhino :D
[20:42:13] <SWPLinux> tutorials maybe
[20:42:32] <SWPLinux> (I may be wrong about rhino BTW - I think I have a several year old version)
[20:42:55] <SWPLinux> so, you make a circle, then make a smaller circle, then do a loft between the two to get a section of a cone
[20:42:57] <ds3> it is attractive given that I get madcam for free and rhino is sub $200 ;)
[20:43:09] <SWPLinux> cadmax is sub $300 :)
[20:43:40] <SWPLinux> in rhino, I don't think you can change the top circle to a square, for example, then redo the loft
[20:43:55] <Dallur> SWPLinux: nope, but you can do things like this http://dallur.com/typo3temp/pics/5c5f60c9b4.png
[20:43:55] <SWPLinux> with a parametric modeler that's sketch-based, you can
[20:44:23] <alex_joni> SWPLinux: there is always alibre :)
[20:44:24] <Dallur> SWPLinux: then you can flatten it and export to dxf, convert to gcode and make your boat :D
[20:44:38] <SWPLinux> Dallur: try changing the profile though
[20:44:45] <ds3> Hmmm
[20:44:55] <SWPLinux> or make it 10 feet (3m) longer
[20:44:58] <ds3> interesting, something I won't really find doing just a plain eval
[20:45:19] <SWPLinux> or extend the rudder ...
[20:45:28] <alex_joni> Alibre Design Express (full parametric modeller) is free :)
[20:45:36] <Dallur> SWPLinux: you can actually extend and stretch things in rhino, with a boat that would be a bad idea though
[20:45:50] <SWPLinux> it's ad/crap/annoyance-ware
[20:45:52] <fenn> alex_joni: zero cost
[20:46:02] <fenn> well, zero monetary cost at least :P
[20:46:03] <alex_joni> fenn: yeah, that
[20:46:15] <ds3> the current version of ALibre Xpress doesn't do ads
[20:46:20] <alex_joni> the full version is 2000$ or so, but features cam
[20:46:26] <alex_joni> ds3: it's not free either
[20:46:33] <ds3> alex_joni: Xpress is
[20:46:37] <alex_joni> free as in free software
[20:46:39] <SWPLinux> Dallur: can you change the profile in the middle of the boat? (for instance, make it a bit flatter/wider), and then re-create the surface?
[20:46:39] <alex_joni> like emc is
[20:46:43] <ds3> oh yeah
[20:46:48] <ds3> neither is Rhino
[20:46:54] <Dallur> SWPLinux: yup, just turn on the control points
[20:47:06] <ds3> problem is I can't figure out alibre quickly like with rhino
[20:47:15] <fenn> so i wonder if vtk would be suitable for use as a cad program's display layer
[20:47:24] <SWPLinux> ok - that allows you to deform existing geometry, but you can't change the base geometry ...
[20:47:27] <alex_joni> ds3: I actually figured alibre way faster out than rhino
[20:47:48] <daniel_br> for rhino lovers http://moi3d.com/
[20:48:10] <ds3> alex_joni: have you used other 3D cads before? I have 0 3D prior experience
[20:48:13] <Dallur> SWPLinux: change the base geometry, you mean extend items, sure you can, you can extend to any other object, by a % or by a fixed length, solids, lines and surfaces
[20:49:38] <SWPLinux> hmmm, so if I make a fillet on two surfaces, then I can go back and just change the radius from 0.2 to 0.5 whenever I want?
[20:49:44] <SWPLinux> s/on/between/
[20:50:20] <SWPLinux> I thought that the fillet became a new entity (deformable, of course), but that you can't go back and recreate it with different parameters
[20:50:44] <Dallur> SWPLinux: Nope, but you can recreate it about as fast as you can change the parameter so ....
[20:50:58] <alex_joni> ds3: I used autocad 3D a bit
[20:51:05] <SWPLinux> except that the parameter can be a formula based on some other thing that I change
[20:51:13] <SWPLinux> (in a real parametric modeler)
[20:51:16] <alex_joni> but alibre was actually the first I really used to design something, and I really liked it
[20:51:34] <alex_joni> Dallur: parametric is all about constraints and formulas
[20:51:48] <alex_joni> make two lines paralel, and vary the distance between them
[20:51:50] <alex_joni> super fast
[20:52:02] <ds3> I'd give alibre another try but it insist on 2K SP1 and the system is not on the network for security reasons :/
[20:52:08] <alex_joni> check it out.. it's "free"
[20:52:22] <alex_joni> ds3: works a treat on XP
[20:52:36] <ds3> alex_joni: no XP licenses around
[20:52:47] <alex_joni> I especially like the feature that you can create pdf's with a 3D view of the design when you're finished
[20:52:55] <SWPLinux> so in CadMax / SolidWorks, I can create a motor mount (for example :) ) by making a sketch that has the bolt pattern and shaft hole, and I can change the location of the features (holes) created by the sketch geometry just by editing one number (a distance to some locator point on the sketch)
[20:53:03] <alex_joni> so you distribute those, and your customer/whoever can rotate it verynicely
[20:53:13] <SWPLinux> I can also change the features by editing the sketch, but leaving the location alone ...
[20:54:09] <SWPLinux> and if I want, I can make the bolt pattern parametric as well, so I can vary the radius of the bolt circle and the number of bolts just by changing a couple of numbers in something that looks like a spreadsheet
[20:54:11] <alex_joni> Dallur: here's a nice boat for you http://www.alibre.com/images/gallery/nz-canoe.jpg
[20:54:24] <daniel_br> Ironcad http://www.ironcad.com/gallery/All is another very good 3d parametric modeler easy for learn but not cheap
[20:55:22] <SWPLinux> do you know how much it costs?
[20:55:42] <daniel_br> 4000 usd
[20:55:47] <daniel_br> i think
[20:55:51] <SWPLinux> ok - a bit less than SolidWorks
[20:57:00] <Martin_Lundstrom> it looks like madcam is swedish :)
[20:59:32] <ds3> yes, the developer is swedish
[20:59:37] <ds3> but the price of madcam is right
[20:59:47] <ds3> ($0.00)
[21:00:03] <Martin_Lundstrom> :)
[21:00:23] <fenn> is it not $640?
[21:00:27] <daniel_br> madcam is 600 euros i think
[21:00:29] <alex_joni> hmm there's ladder in ironcad
[21:00:34] <ds3> their competive upgrade program
[21:00:44] <alex_joni> http://www.ironcad.com/gallery/IndustrialDesign/Quick_Step_snett_ovanifr_n_2003_01_19
[21:00:55] <ds3> send a copy of the manual for a CNC machine, other CAM/CAD package and get MadCAM 2.0 for free
[21:01:05] <ds3> so $0.00 =)
[21:01:20] <alex_joni> * alex_joni wanders off to write a manual
[21:01:22] <SWPLinux> wow - they have quickstep and ladder in there :)
[21:01:36] <alex_joni> SWPLinux: my points exactly :D
[21:01:39] <fenn> GPL violation sue sue!
[21:01:45] <SWPLinux> copyright
[21:01:49] <SWPLinux> err - left
[21:01:51] <SWPLinux> err - left right left
[21:02:00] <fenn> attention!
[21:02:19] <SWPLinux> more like "ten-hut!"
[21:02:23] <SWPLinux> or something
[21:02:39] <fenn> * fenn emits a loud barking noise
[21:02:47] <SWPLinux> exactly that
[21:03:08] <alex_joni> this is nice: http://www.ironcad.com/gallery/IndustrialDesign/bottom1345?full=1
[21:03:32] <fenn> yeah i wonder who trusts 3d models of electronic parts
[21:03:58] <SWPLinux> well, Altium lets you import IGES models into the part libraries, and you can export an IGES of the whole board from it
[21:04:29] <fenn> kicad does 3d models :)
[21:04:52] <fenn> does iges include color info?
[21:04:52] <SWPLinux> you can bet that the part manufacturers have pretty accurate models of their parts
[21:04:57] <fenn> like textures etc
[21:05:01] <SWPLinux> I don't know
[21:05:15] <SWPLinux> I think there are other imports as well, but I don't recall them offhand
[21:05:36] <alex_joni> eagle has an add-on for 3d rendering of designs
[21:05:48] <fenn> its pov ray though
[21:05:54] <fenn> what's the point of that?
[21:06:17] <SWPLinux> the cool thing with Altium is that it can actually import real part models, not just create things that are more or less thick blobs ...
[21:06:20] <alex_joni> fenn: see the design before it's built?
[21:06:22] <ds3> do any of them generate a bounding box for the entire board?
[21:06:34] <SWPLinux> possibly
[21:06:56] <SWPLinux> the main places where you'd use a true 3D model in mechanical CAD wouldn't really need a bounding box though
[21:07:07] <SWPLinux> they'd need a lot more detailed info
[21:07:11] <SWPLinux> (like a model :) )
[21:07:30] <ds3> I'm interested in knowing what kind of standaard case I'd need to house it
[21:07:43] <ds3> or if it will slide in a card cage w/o consuming multiple slots
[21:07:47] <fenn> well, find your tallest component and the board dimensions..
[21:07:57] <SWPLinux> right - what he said
[21:08:01] <ds3> i know how to do it by hand
[21:08:27] <ds3> but be nice if the software can have it as part of the library
[21:08:38] <daniel_br> the best free 3d library http://www.3dcontentcentral.com/3dcontentcentral/cc_frame.asp
[21:08:48] <SWPLinux> oh- it can take STEP models
[21:08:51] <fenn> "the best"?
[21:09:07] <daniel_br> imo
[21:09:28] <SWPLinux> if you can view flash files, look here: http://www.altium.com/Evaluate/DEMOcenter/WhatsnewinAltiumDesigner/ImprovedStepModelImportFeature/
[21:14:00] <alex_joni> SWPLinux: is there anything altium doesn't do?
[21:14:41] <skunkworks> I doubt if it gets you coffee and donuts
[21:15:43] <cradek> mmm coffee
[21:32:39] <alex_joni> eek Server Error
[21:32:40] <alex_joni> We're sorry, but Gmail is temporarily unavailable. We're currently working to fix the problem -- please try logging in to your account in a few minutes.
[21:33:01] <SWPLinux> alex_joni: yes - it doesn't do Linux :(
[21:33:16] <alex_joni> SWPLinux: no, this is on doze
[21:33:28] <alex_joni> works now.. it was a glitch
[21:33:42] <SWPLinux> that's Altium that doesn't do Linux
[21:33:46] <alex_joni> ahh.. ok
[21:34:09] <SWPLinux> that's why I have a copy of XP64 Pro, and may end up buying VMWare or XenWorks
[21:34:11] <alex_joni> well.. at 10k$ license I think you can afford 200$ for doze
[21:34:24] <alex_joni> for the license I mean
[21:34:28] <SWPLinux> it's the extra computer that bothers me ;)
[21:34:40] <alex_joni> Xen
[21:34:45] <SWPLinux> and it's $12k with the FPGA stuff;)
[21:34:56] <alex_joni> 10k, 12k whatever :D
[21:35:29] <SWPLinux> I was looking at Xen things recently. it looks like they're pretty close to bare metal speed with the current crop of CPUs
[21:35:47] <alex_joni> nice
[21:35:54] <SWPLinux> I haven't gotten the beta of VMWare 6 to run yet, so I havent tested that yet
[21:40:34] <anonimasu> hm
[21:40:36] <anonimasu> hi btw..
[21:42:20] <awallin> anyone know what RS422/485 is ?
[21:42:25] <SWPLinux> yes
[21:42:28] <anonimasu> yes?
[21:42:33] <SWPLinux> yes. :)
[21:42:36] <cradek> yes
[21:42:36] <awallin> care to elaborate?
[21:42:44] <SWPLinux> RS422 and RS485 are signaling protocols for serial data
[21:43:00] <anonimasu> actually http://en.wikipedia.org/wiki/EIA-422
[21:43:01] <lerman> jepler: or anyone having experience with images on Buttons in Tkinter. How? (I have a png image.)
[21:43:12] <SWPLinux> they use a differential pair (I don't recall the actual voltage thresholds) for each signal
[21:43:34] <anonimasu> I think you can do multi devices on rs422.. if I remember it right
[21:43:35] <SWPLinux> one of them (I think RS-422) is point-to-point, like RS-232
[21:43:58] <SWPLinux> the other (I think RS-485) is point-to-multipoint, which allows multiple devices to sit on the bus
[21:44:03] <anonimasu> no..
[21:44:06] <anonimasu> that's 485..
[21:44:13] <awallin> so, if I have a normal serial port in my pc (thats rs-232) can I do rs422 easily?
[21:44:16] <SWPLinux> I just said that
[21:44:20] <alex_joni> awallin: with an adapter
[21:44:47] <SWPLinux> you need a level converter, and RS-4{22,85} needs extra care if you want multiple devices
[21:45:19] <SWPLinux> http://www.rs485.com/rs485spec.html
[21:48:49] <awallin> rs422 would be for controlling a vector vfd
[21:49:07] <SWPLinux> and 485 for several servo controllers
[21:49:17] <awallin> now we have a cheap V/f one, but it's crap at low rpm
[21:50:09] <anonimasu> I wish I had a vfd :)
[21:50:29] <anonimasu> I hate the pricing on big ones
[21:50:45] <SWPLinux> ebay can be good for that
[21:50:56] <SWPLinux> though I haven't gotten anything larger than 3 HP
[21:51:28] <anonimasu> I need a 5kw
[21:51:29] <awallin> now that I am actually looking at the datasheet, it seems that it also has a potentiometer, or pulse train input for setting the speed
[21:51:30] <anonimasu> one :/
[21:51:46] <anonimasu> they are around 800 eur new
[21:51:49] <anonimasu> the cheap ones
[21:51:53] <SWPLinux> yeah - that's a problem
[21:52:58] <awallin> could probably filter a pwm output from the m5i20 and use that as analog control for the vfd
[21:56:36] <SWPLinux> anonimasu: http://web1.automationdirect.com/adc/Shopping/Catalog/AC_Drives_-z-_Motors/GS2_(115_-z-_230_-z-_460_-z-_575_VAC_V-z-Hz_Control)
[21:56:55] <SWPLinux> they may even ship there (or have a European counterpart - they have one in .au)
[21:58:36] <anonimasu> hm
[21:58:39] <anonimasu> dosent seem to open
[21:59:02] <SWPLinux> hmmm. try http://www.automationdirect.com and navigate to it
[21:59:15] <anonimasu> hm yeah now
[22:01:15] <fenn> those are a lot cheaper than i thought
[22:01:40] <SWPLinux> yeah - they've been surprisingly low for some stuff
[22:01:53] <SWPLinux> stepper motors, for example (276 oz-in for $39)
[22:02:17] <SWPLinux> drives and servos, however, are at least as pricey as expected
[22:02:44] <anonimasu> hm ok
[22:02:44] <anonimasu> yeah
[22:02:51] <anonimasu> maybe I should order one later
[22:05:47] <alex_joni> good night all
[22:05:54] <SWPLinux> see you Alex
[22:06:08] <anonimasu> night alex
[22:08:26] <lerneaen_hydra> 'night
[22:16:37] <K`zan> My belt drive may arrive today! Making the most sense out of USPS tracking that is...
[22:17:02] <tomp> nite alex
[22:17:03] <K`zan> SWPLinux: You back already? Wow, such diligence :).
[22:17:08] <SWPLinux> heh
[22:17:17] <SWPLinux> been here since 10-ish :)
[22:17:31] <K`zan> So, if I may ask, being courious, what is it that you do sell?
[22:17:34] <SWPLinux> I suppose I should do some real work though (helps with the mortgage)
[22:17:51] <SWPLinux> I sell my programming and sometimes electrical engineering services
[22:17:54] <K`zan> I've been up since 0900, but daily tasks just reached a break point for a while.
[22:18:15] <SWPLinux> I may one day sell some products, but as yet I don't have any that I'm actively trying to sell
[22:18:31] <K`zan> SWPLinux: Cool, love to program, but am not going to get my arse in another 18-20 hour day thing with clueless management :-(.
[22:18:51] <SWPLinux> I own my business, so clueless management isn't an issue ;)
[22:18:53] <K`zan> Someday I will finish up my anamatronic controller, that should sell.
[22:19:21] <tomp> i read JMK posted something about a DRO and pyvcp, been checking mail all day, cant find it, maybe another group?
[22:19:26] <K`zan> Yep, that is what I want to try to get to. One of the main reasons we are moving - no business out of the apartment allowed (and pretty much anything else here).
[22:20:15] <anonimasu> K`zan: I hope you tell the owner of the house to fsck off.. when you leave..
[22:20:17] <jepler> tomp:
[22:20:17] <jepler> From: [email protected]
[22:20:17] <jepler> To: "Enhanced Machine Controller (EMC)" <[email protected]>
[22:20:17] <jepler> Date: Thu, 04 Jan 2007 18:21:33 +0000
[22:20:17] <jepler> Subject: Re: [Emc-users] Qt3 lib and dro
[22:20:39] <jepler> ^^^ ths is the only one I found in my mail archive
[22:20:42] <anonimasu> ^_^
[22:20:47] <K`zan> anonimasu: I already have a 5 page review that I am going to put up on the apartment reviews site when I leave...
[22:20:50] <jepler> bbl
[22:21:13] <ejholmgren> K'zan: just burn it down
[22:21:30] <tomp> jepler: thanks, didnt appear in my mail & isnt on sourceforge emc-users, even tho later ones ( ie "rc servo" ) are there
[22:21:32] <K`zan> The owner is a hitler youth that got imported here after WWII and went to work for boeing. Freely admits to being a "key nazi" :)
[22:21:56] <K`zan> Not the worst slum we ever lived in, but close :-/.
[22:22:19] <K`zan> BIG improvement over the last one :-).
[22:23:03] <K`zan> ejholmgren: Nah, that would get him out of it - he paid WAY too much for this place under the guise of "Get Rich In Real Estate", I want him to suffer :).
[22:24:00] <K`zan> Really looking forward to getting that belt drive :-)!
[22:24:13] <anonimasu> ^_^
[22:24:24] <ejholmgren> just drop dead animals down anything open on the roof then :)
[22:24:42] <K`zan> :-), I thought of that or releasing a termite colony :)
[22:24:49] <anonimasu> mm
[22:24:51] <anonimasu> thermite is nice
[22:25:12] <K`zan> Just wish some of these 1970s appliences would die...
[22:25:23] <K`zan> correction: finish dying...
[22:25:27] <anonimasu> ;)
[22:25:29] <anonimasu> brb
[22:25:33] <anonimasu> going to have a nice bath
[22:25:36] <Dallur> Just leave the fridge open and the stove on for a couple of days
[22:26:07] <K`zan> fridge runs allmost all the time anyway, eyes on the stove are shot so we do most of our cooking on the grill out on the patio.
[22:26:35] <K`zan> microwave was a real goodie in 1970, but is now down to 2.5 minutes to get a cup of coffee hot.
[22:27:06] <K`zan> But electricity is cheap here, our water bill is as much or more than that.
[22:27:20] <ejholmgren> K'zan: wherabouts?
[22:27:36] <K`zan> scabattle (seattle WA USA).
[22:28:14] <K`zan> Taxes are skyrocketing here and that is really hurting him...
[22:28:39] <K`zan> Taxes went up $30K last year (28 unit building)...
[22:29:06] <K`zan> Will be SO glad to get out of this socialist shithole.
[22:30:37] <lerneaen_hydra> heh, whatever you do, don't move to sweden ;)
[22:30:55] <K`zan> No worries on that :-), I don't speak arabic :)
[22:31:16] <lerneaen_hydra> err, surely you jest
[22:31:38] <K`zan> Actually, I think Sweden has gotten socialism to work far better than most, if what I read is correct.
[22:31:59] <lerneaen_hydra> still, it's not exactly working very well IMO
[22:32:17] <K`zan> Socialism, for some reason, never does...
[22:32:27] <ejholmgren> http://www.hamiltonspectator.com/NASApp/cs/ContentServer?pagename=hamilton/Layout/Article_Type1&c=Article&cid=1168470616997&call_pageid=1020420665036&col=1014656511815
[22:32:30] <K`zan> Great idea, but it just never seems to work.
[22:32:34] <ejholmgren> have to love crazy inventors
[22:33:28] <K`zan> Yep, where can I get one of those :-)?
[22:37:10] <ejholmgren> K'zan: you should move to MN where they tax 1/3 of your income
[22:37:19] <ejholmgren> commie bastards
[22:37:42] <K`zan> Directly, here they hide it all in a million things that get taxed several times as it comes to you.
[22:37:53] <ejholmgren> my paycheck looks so much nicer when I read the first line
[22:38:34] <K`zan> we really need to implement the "rope, politician, lampost - some assembly required" idea, as long as they can pick you pocket without penalty we are indeed screwed...
[22:39:11] <K`zan> But people keep voting for dimocrats, never figured that one out. Not that the RINOs are significantly better these days :-(.
[22:39:43] <lerneaen_hydra> you're complaining about 1/3 income tax?
[22:39:44] <lerneaen_hydra> heh
[22:39:56] <lerneaen_hydra> do not come to sweden, other than for a vacation
[22:40:04] <K`zan> Pols can never get enough $$$ to squander.
[22:40:24] <K`zan> Sadly, I make it a point to stay out of europe anymore :-(.
[22:40:45] <K`zan> Seattle is bad enough :-).
[22:41:41] <lerneaen_hydra> youre from seattle?
[22:42:18] <K`zan> No, I ended up here on a job and liked the physical environment, but the rest of it just isn't worth it anymore.
[22:42:28] <lerneaen_hydra> oh, I see
[22:43:08] <K`zan> I thought chicago was crooked and corrupt, seattle (western WA) has them as babes in the woods :-(.
[22:43:51] <lerneaen_hydra> huh?
[22:44:50] <K`zan> chicago could learn a lot about graft, greed and corruption from seattle...
[22:45:09] <lerneaen_hydra> oh, right
[22:45:15] <K`zan> Sadly so.
[22:46:26] <K`zan> We are hoping to move to north central WA, 99.995% due to the physical environment.
[22:46:37] <lerneaen_hydra> I only lived in seattle when I was a kid, so I only remember the good bits
[22:47:00] <K`zan> It was massively better 10 years ago or so when I first came out here.
[22:47:16] <lerneaen_hydra> oh, I was there ~10 years ago too
[22:47:19] <K`zan> No comparison to what it is now.
[22:47:24] <lerneaen_hydra> haven't been there since
[22:47:33] <lerneaen_hydra> damn, pity to hear it's worsened
[22:48:02] <K`zan> Yes, it really is, this area has the nicest physical environment I've experienced.
[22:48:21] <K`zan> When I came here, people would let you out in traffic - not any more...
[22:48:35] <K`zan> Makes chicago look polite :).
[22:49:00] <lerneaen_hydra> hmm, let you out in traffic? out from the sidewalk?
[22:49:07] <K`zan> sidestreets.....
[22:49:35] <K`zan> About a week of freezing weather now and the side streets are still skating rings...
[22:49:36] <jepler> crap crap -- the machine I want to put the mesa in has a "screwless" card retention system which interferes with the top of the m5i20 board
[22:49:58] <K`zan> bugger it with a screw as normal?
[22:50:00] <jepler> (instead of a screw, a wide plastic tab is pressed down over the card bracket)
[22:50:07] <K`zan> Can't remove it?
[22:50:13] <cradek> jepler: I've used sidecutters on those stupid plastic tabs
[22:50:15] <jepler> not without breaking it
[22:50:25] <K`zan> did that on this case, great idea, but it doesn't work well IMO.
[22:50:25] <jepler> not that breaking it doesn't sound attractive
[22:50:34] <cradek> yeah, that's the general idea
[22:50:45] <K`zan> Trust me with the new form factors, it is not appreciating in value :).
[22:51:39] <K`zan> This one had that for the drive bays too, fortunately you could just remove those.
[22:52:46] <SWPLinux> is there still a screw hole there?
[22:52:54] <jepler> argh -- out of ethernet cables
[22:53:09] <SWPLinux> how can that happen?
[22:53:49] <K`zan> Yes, these were added on to a regular drive bay, they work, but not to my satisfaction, to begin with you loose the heat sinking since they more or less sorts sit in the bay.
[22:55:03] <jepler> found one!
[22:55:10] <K`zan> :-)!
[22:55:38] <K`zan> God knows what I'll find I have for cables if I ever unbraid this mess behind everything :).
[22:55:53] <anonimasu> iab
[22:56:17] <K`zan> My theory: Cables routed out of sight braid themselves together in the night.
[22:56:21] <anonimasu> K`zan: that's surprising as you live in usa..
[22:56:28] <anonimasu> K`zan: compared to it sweden is nice
[22:56:30] <K`zan> anonimasu: Huh?
[22:56:45] <K`zan> Yes, def, it could be far worse than it is.
[22:56:56] <K`zan> But it is working hard at getting to that.
[22:57:00] <anonimasu> who is?
[22:57:25] <K`zan> WA / Seattle / King County governments.
[22:58:00] <jepler> hard drives have a LOT of metal and don't use more than a few watts -- I am surprised they would need additional heatsinking
[22:58:32] <K`zan> jepler: Some of them do get toasty, just always better cool than hot.
[22:58:53] <jepler> my latest case has rubber washers to hold the HD off the metal, with the purpose being to reduce noise
[22:59:01] <anonimasu> ah
[22:59:14] <anonimasu> jepler: if you stack them they get really hot..
[22:59:19] <anonimasu> jepler: like in older computer cases :)
[22:59:24] <K`zan> jepler: Heh, I could stand the noise reduction and the heat with as cold as it is would not be unwelcome either :).
[22:59:51] <K`zan> We are supposed to get more snow, unheard of here...
[22:59:56] <jepler> anonimasu: I think there's probably at least 1/4" air between two drives .. but I only have one HDD at the moment
[23:00:16] <K`zan> I remember my stacks of 40M drives :-) LOL...
[23:01:05] <jepler> indeed
[23:01:48] <jepler> compare that to the ease with which we now have 1TB of disk space spread over 3 or 4 machines...
[23:02:32] <K`zan> about that here with all the drives. Amazing isn't it :-).
[23:03:22] <K`zan> 320G, 200G, 4-80G, a few 40s in the closet, a 2G 5.25 and an early 1.2G :-).
[23:03:31] <Dallur> What is really boggles me is that now you can get network storage with 3TB for less than $1500
[23:03:57] <K`zan> Yep, I was pleased as hell to get my 300M drive for about $500 not all that long ago now.
[23:04:14] <K`zan> full high 5.25 :-).
[23:04:39] <K`zan> attached to a 386sx20 :).
[23:04:49] <Dallur> lol, old scsi one with the 25 pin connectors that looked like printer plugs :P
[23:05:08] <K`zan> All those went here when I dumped all the Macs...
[23:05:17] <SWPLinux> lacie has some (refurb) 1TB USB drives for $299
[23:05:26] <K`zan> LUST!!!!!!!!!!!!!!!
[23:05:42] <jepler> halcmd: loadrt hal_m5i20
[23:05:43] <jepler> whee
[23:05:44] <jepler> it works
[23:05:49] <K`zan> Actually, even with as sloppy a housekeeper as I am the 320G drive is still pretty empty :).
[23:05:51] <SWPLinux> yay!
[23:05:53] <jepler> * jepler wanders off again
[23:06:20] <SWPLinux> heh - I've got 2x320G, 2x300G and 5x120G (plus some smaller ones)
[23:06:20] <jepler> SWPLinux: 1TB? wow.
[23:06:27] <SWPLinux> oh right - the 3x36G SCSI RAID also
[23:06:27] <K`zan> Was really happy with the DVD-RW until it dawned on me that those are the equivalent to floppies for the drives we have now :-).
[23:06:34] <jepler> I've got: 320GB, 300GB, 250GB, 250GB, 80GB, and they get smaller from there
[23:06:48] <jepler> the smallest might be 850MB or 4GB, not even sure
[23:06:51] <SWPLinux> http://www.lacie.com/products/clearance/
[23:08:34] <K`zan> * K`zan tells self: while (1) printf("N E E D C N C M O R E !!!n);
[23:08:43] <K`zan> :-)
[23:09:35] <K`zan> We use the old 2G 5.25 as a power supply load for Jeans electroplating PC power supply :).
[23:10:10] <K`zan> * K`zan busily NOT following that link :).
[23:10:59] <SWPLinux> I like the external CD/DVD duplicator for $549 (one master -> 5 writers simultaneously)
[23:11:15] <K`zan> That might be a good start with double density DVDs :-).
[23:28:36] <robin_sz> mweep?
[23:31:15] <fenn> mmm vtk is damned sexy
[23:32:16] <K`zan> wb SWPadnos
[23:32:23] <SWPadnos> thans
[23:32:24] <SWPadnos> ks
[23:32:26] <SWPadnos> thanks
[23:32:32] <K`zan> yo
[23:32:33] <K`zan> u
[23:32:36] <K`zan> are
[23:32:39] <K`zan> welcome
[23:32:41] <K`zan> :-)
[23:32:44] <SWPadnos> :)
[23:32:45] <K`zan> LOL