#emc | Logs for 2009-08-20

Back
[02:36:35] <jepler> whee, I finally made a circuit board on the zenbot again -- the first once since I converted it to servos
[02:37:11] <jepler> some itty bitty boards that will let me put the home switches on the unused index inputs of the 5i20
[02:37:48] <jepler> It's the first board I did without using the tool length sensor. I just relied on the depth-setting rings to give me a repeatable tool length. Touched off with the critical (trace isolation) tool. That worked just fine
[02:37:53] <mozmck> neat! I need to get a little router built to do stuff like that.
[02:38:36] <mozmck> what is the trace isolation tool?
[02:38:54] <jepler> It's a V-bit tool. It's what cuts the copper, isolating each trace from the others
[02:39:37] <jepler> these are the tools I use: http://precisebits.com/applications/pcbtools.htm
[02:39:37] <mozmck> ok. did you touch off manually then by moving down until the tip touched the board?
[02:40:44] <jepler> I use the dowel-pin method with a broken 1/8" shank tool instead of a dowel pin
[02:41:11] <mozmck> guess I haven't heard of that method.
[02:41:26] <mozmck> does the router run faster with the servos?
[02:41:35] <jepler> I jog down until I'm 1/8 minus a bit above the copper, then jog incrementally up until the pin rolls freely under the tool, then jog down again until it just catches
[02:41:48] <jepler> then I enter the pin's diameter in touch-off, 1/8 or .125
[02:42:38] <mozmck> ah, I see. that would be more accurate than the method I use, but the stuff I've done so far on my router doesn't need any accuracy on the Z
[02:44:11] <jepler> I can go much faster without ferror than I could go without stalling the old steppers (2.5+ inch/second vs 1.4inch/second), but the machine makes a lot of noise above 50inch/minute or so
[02:46:51] <mozmck> That's probably pretty good for that machine though, being so small.
[02:47:05] <jepler> most time is spent cutting, not doing rapids
[02:47:30] <mozmck> yeah, and that is probably much slower.
[02:48:27] <mozmck> good night
[02:48:30] <jepler> see you
[02:56:02] <Valen> what spindle speed you running
[02:56:02] <Valen> ?
[03:00:35] <jepler> I'm not sure precisely. It's supposed to be around 20k.
[03:00:55] <cradek> jepler: how does the pcb look? sharp corners?
[03:02:15] <jepler> what I'm puzzled by is how far off-center the drills look to be
[03:02:28] <cradek> hm
[03:02:47] <cradek> do you still do a spot drill with the v tool?
[03:03:00] <cradek> not that it should be needed with real pcb drills
[03:03:29] <SWPadnos> cradek, I think I might finally have a chance to send you a few Aluminum tubes for lathe work. Do you think you have time to mess with them in the next week or so?
[03:04:00] <cradek> I didn't know you were serious - what do they need?
[03:04:07] <SWPadnos> heh
[03:05:03] <SWPadnos> well, the overall job would be a light cut on the OD (to make it look nice, not for an exact measurement), then drill or bore and thread the inside of both ends, with different diameters of course (same pitch though)
[03:05:16] <jepler> http://emergent.unpy.net/files/sandbox/img_5795-medium.jpg
[03:05:25] <SWPadnos> I can drill with a nearly-wide-enough drill to make the boring faster, and I can tap one end
[03:05:51] <SWPadnos> oh, and facing it off square on both ends would be nice, since I don't do the best job with a hacksaw :)
[03:06:34] <cradek> I can't easily face the whole OD in my chucker, but the rest sounds like something I can do
[03:06:57] <cradek> what are the inside threads?
[03:07:16] <SWPadnos> 1mm pitch, I think one is 20.5mm, the other 18mm (which I now have the tap for)
[03:08:17] <cradek> 18mm I can still single-point
[03:08:31] <cradek> but I do not have a good way to measure inside PD
[03:08:46] <cradek> is this cut-to-fit-something or does it need to be good?
[03:09:04] <SWPadnos> well, it
[03:09:17] <SWPadnos> it's supposed to match a particular XLR connector I have
[03:09:38] <SWPadnos> which I can send you
[03:09:46] <SWPadnos> the thread is 20.5x1mm
[03:10:33] <cradek> can you send a sketch? how many of these do you need? (starting to sound like more than a couple hours of work)
[03:10:47] <SWPadnos> yeah, I can sketch it
[03:11:14] <SWPadnos> and do a lot of it too, it's just that 20.5mm thread that gets me down, and the OD turn isn't strictly necessary - it's just cosmetic
[03:11:30] <SWPadnos> getting rid of the mill finish and markings
[03:11:44] <SWPadnos> now I wish I had bought that bead-blaster :)
[03:12:55] <jepler> copper clad sure looks weird up close -- grainy and multicolored. http://emergent.unpy.net/files/sandbox/img_5795-crop.jpg
[03:13:15] <cradek> seems like setting one tool will give both threads without much added effort
[03:13:16] <SWPadnos> I had never noticed that pixelated look before
[03:13:35] <SWPadnos> yeah, though it is too bad they're on oposite sides
[03:13:51] <jepler> it's not sensor noise -- it doesn't appear in the dark area at the bottom of the hole
[03:13:55] <SWPadnos> actually, I have only sawed one length in half - it was bought as two 1 foot tubes
[03:14:15] <cradek> sawing is trivial
[03:14:28] <cradek> don't bother if all you have is a hacksaw
[03:14:30] <cradek> (ugh)
[03:14:46] <SWPadnos> well, I was thinking that the overall length isn't important - I only used 6" since I can easily saw it in half
[03:15:04] <SWPadnos> well, I could fire up the Johnson, but it seems like overkill for a 1" OD aluminum tube
[03:15:35] <jymm> chainsaw
[03:15:41] <cradek> jepler: those corners look good. you must have high accel.
[03:15:47] <jymm> angle grinder
[03:15:48] <SWPadnos> how much material do you think you'd need in the chuck? mayne 1/2" ro 1"?
[03:16:06] <cradek> I can thread inside the collet if I need to
[03:16:21] <cradek> the part should probably be at least an inch long
[03:16:50] <SWPadnos> yeah, for the threading you don't need much - I was thinking that the OD turn could be done on the long bar (12 or 6"), then cut off what's in the chuck
[03:17:10] <cradek> yeah, depends how long you need the part to be
[03:17:11] <SWPadnos> the overall length should be somewhere between 4.5 and 6 inches. 5 feels pretty good
[03:17:45] <skunkworks928> jepler: Cool!
[03:17:47] <cradek> 5 is probably too long to turn the OD
[03:18:03] <jepler> cradek: yeah that part is nice
[03:18:07] <jepler> the drills, not so much
[03:18:09] <SWPadnos> 4.25 wouild be too short, 6 seems too long, so somewhere in between :)
[03:18:15] <cradek> only one looks badly off center...
[03:18:37] <cradek> 5 +- 0.5 is a pretty easy tolerance to hit
[03:18:44] <jepler> I think the accel is just a hair higher than my old stepper config -- 32 vs 30 in/sec2
[03:19:33] <cradek> even the bad one looks good enough - maybe imperfect servo tuning?
[03:19:52] <cradek> assuming it's not just backlash (big assumption)
[03:20:13] <jepler> MIN_FERROR=.001 on X and Y, so it seems like it should be closer than that
[03:20:19] <cradek> oh yep
[03:20:25] <cradek> maybe the board moved
[03:20:45] <cradek> you didn't drill after cutout did you?
[03:21:22] <cradek> have you checked position repeatability (no encoder counts messed up due to noise)?
[03:21:35] <jepler> drill before cutout
[03:22:47] <cradek> SWPadnos: I think I would be able to do these no later than the weekend of the 29th
[03:22:50] <jepler> I wouldn't say I've ruled out lost/gained encoder counts
[03:22:53] <skunkworks928> torquing something when changing the bit..
[03:23:22] <skunkworks928> (assuming wrenches required?)
[03:23:29] <jepler> earlier I did a test where I ran the spindle power right next to the encoders with the motors stationary and amps off. I didn't see counts, even single counts, in that situation
[03:23:29] <SWPadnos> cradek, ok. I'll see about doing most of the work here, and maybe I'll search for an M20.5x1 tap too (har har)
[03:23:55] <jepler> early on without the spindle running I did "run an integer number of turns" tests and checked by eye
[03:24:00] <cradek> heh, that sounds fun
[03:24:18] <jepler> wow, this image is nearly 10000dpi
[03:24:47] <SWPadnos> man. time for bed for me - night guys
[03:25:03] <skunkworks928> yah - what is jepler doing up so late?
[03:25:58] <skunkworks928> time for bed here - night
[03:26:30] <jepler> stupid gimp
[03:26:57] <skunkworks928> btw - the puma arm vismach that alex did is really fun to play with.
[03:27:02] <jepler> it has configurable units, but you can't configure one that is less than .005 inches
[03:29:45] <jepler> I must have been cutting way too deep. the traces are 24 mil in eagle but I measure 18
[03:30:09] <jepler> the hole is off center by about 4 mil
[03:32:05] <jepler> and with that, goodnight!
[03:32:05] <cradek> hm, probably not following, probably not backlash
[03:32:25] <cradek> the spindle moved when you tightened it, or the board moved, or you lost position
[03:33:27] <cradek> hey I should go to bed too, tomorrow-me will appreciate it
[07:17:21] <numen> good morning
[08:02:29] <numen> can anyone tell me, what a kind of miller i should use for polystyrol?
[08:26:20] <jymm> a wire
[08:32:41] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[08:32:51] <jymm> SWPadnos: howdy
[09:19:00] <piasdom> g'mornin all
[09:24:11] <numen> wire?
[10:14:58] <archivist> ballscrew arrives from China :))
[10:24:57] <celeron55> do you know what is the cheapest place to get milling cutters for milling pcb's in europe? (more precisely, finland)
[10:25:51] <celeron55> i found this but they have 27 euros as their minimum shipping cost http://www.frezycnc.eu/
[10:34:01] <archivist> someone pointed at engraving cutters a few months ago if you feel like googling the logs
[10:34:24] <celeron55> i guess i'll first try with some discount dremel bits.. when i get my mill working again some day
[10:34:44] <micges_work> celeron55: heh company from poland ;)
[10:35:28] <micges_work> celeron55: what size do you want?
[10:39:16] <alex_joni> celeron55: I got mine from a company in austria, no idea what shipping to finland would be
[10:40:47] <alex_joni> celeron55: http://www.thecooltool.com/files/dateien_533.pdf
[10:47:49] <archivist> celeron55, make your own http://www.pilotltd.net/engraving.htm you need a narrower point though
[10:48:42] <archivist> I have made them here for brass engraving
[10:48:51] <alex_joni> The webpage at http://www.pilotltd.net/engraving.htm might be temporarily down or it may have moved permanently to a new web address.
[10:49:04] <alex_joni> doesn't open from here :/
[10:49:21] <archivist> thats steve blackmores site
[10:49:34] <archivist> I just googled to find it
[10:50:21] <numen> celeron55 try bungard
[10:50:32] <numen> they are about 5euro each
[10:54:33] <numen> http://www.bungard.de/content/view/30/81/lang,english/
[11:21:35] <numen> is it possible, to drill a hss driller with a vhm?
[11:30:04] <archivist> vhm?
[11:33:02] <numen> dont know in english
[11:33:10] <numen> in german its voll hartmetall
[11:33:11] <alex_joni> translate.google.com
[11:33:20] <numen> so the next better one than hss
[11:33:34] <alex_joni> carbide
[11:34:57] <numen> possible
[11:35:22] <numen> yes carbide
[11:35:52] <archivist> never tried, most would probably use edm
[11:36:15] <numen> i dont have edm here :(
[11:49:34] <numen> archivist could it work or not?
[11:53:58] <archivist> what part of never tried did you not understand, but I bet you will break something if you do
[11:54:12] <holy> How can I ask logger_emc for a current log-file?
[11:54:42] <archivist> logger_emc: bookmark
[11:54:42] <archivist> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2009-08-20.txt
[11:55:15] <holy> I wrote "link" instead of the "bookmark"...
[12:05:58] <holy> I wanted to use another (X3,Z2) home position in ~/configs/sim/lathe.ini (just to show people some more 'real' version of the Lathe) -- and each time I used some non-zero X-value, ´╗┐there were an error "can not home when limit switch is reached". Should I change something in corresponding .hal file(s) to be able to put home switcher not at the origin of the coordinates.
[12:13:16] <jepler> holy: sim/lathe.ini emulates a machine where the home switches are both on a single input pin. If you change the home position of X so that it is in the area where the switch is closed, homing won't work. You could also move the Xhomeswpos in simulated_home.hal, or make the switch unshared by removing the HOME_IS_SHARED lines in the hal file and changing this block in simulated_home.hal:
[12:13:22] <jepler> net Xhomesw => or2.0.in0
[12:13:22] <holy> ok
[12:13:24] <jepler> net Zhomesw => or2.0.in1
[12:13:27] <jepler> net XZhomesw or2.0.out => axis.0.home-sw-in axis.2.home-sw-in
[12:13:28] <jepler> instead you'd have net Xhomesw or2.0.in0 => axis.0.home-sw-in
[12:13:37] <jepler> and similar for Z / axis.2
[12:13:44] <jepler> the third line becomes unused and would be removed
[12:14:49] <holy> I will think about it after a dinner (tomato salad, steamed potato, water melon)
[12:15:35] <jepler> you can refer to this to see exactly what I meant: http://emergent.unpy.net/files/sandbox/lathe-home-changes.diff
[12:16:03] <jepler> (- represents old lines, + represents new lines)
[12:16:10] <holy> OK, thanks.
[12:16:31] <jepler> bbl
[12:16:44] <christel> [Global Notice] Hi all, I'm about to do some re-hubbing and take some servers down in preparation for todays maintenance, it will be somewhat noisy as I swap things around but shouldn't take particularly long. Apologies for the inconvenience, thank you for flying freenode and have a great day!
[12:28:20] <holy> It's probably called a "net split"...
[12:29:30] <holy> Use "irc.ubuntu.com".
[12:30:26] <holy> Weird times began...
[12:31:38] <numen> is it possible, to drill hss with carbide?
[12:31:53] <holy> What is "hss"?
[12:32:25] <mikegg> high speed steel
[12:32:28] <mikegg> I think so
[12:32:44] <numen> i broke a center drill and this i have to remove
[12:32:48] <mikegg> I've had marginal success drilling out carbide taps with carbide drills
[12:32:54] <mikegg> so...
[12:32:58] <numen> but i have just a 3mm carbide mill
[12:33:14] <mikegg> how fast is your spindle?
[12:33:24] <numen> i think about 18k rpm
[12:33:29] <numen> but no cooling
[12:33:56] <mikegg> just get a spray bottle
[12:34:03] <numen> filled with?
[12:34:16] <mikegg> could you drill from the oposite side and then knock it out?
[12:34:25] <numen> nope
[12:35:40] <mikegg> windex.. whatever you got lying around..
[12:36:58] <holy> numen: 18K revolutions... ***You probabaly*** should read about the "g21 g95 f0.2" feed rates
[12:37:56] <holy> guaranteed advance, though a little one, but with a huge cutting speed.
[12:38:46] <archivist> I use diamond burs for that job
[12:40:48] <numen> but cooling with what?
[12:42:48] <holy> numen: there are many advices on the coolants... Use the machinery's hadbook
[12:43:15] <archivist> carbide often is used without
[12:43:55] <piasdom> archivist:you have a picture of a diamond burr ?
[12:44:01] <archivist> but its brittle so be careful
[12:44:24] <holy> archivist: these "carbide" drils are made of a high-in-carbon-content steel?
[12:44:51] <piasdom> archivist: didn't know i could use that to remove bits and c-drills
[12:46:17] <archivist> http://www.amazon.com/Diamond-Burr-Set-120-Grit/dp/B0012DLA3U
[12:46:49] <piasdom> archivist: Thanks
[12:47:04] <archivist> cheap and dispoable for that sort of job, I use a hand held dremel
[12:50:15] <archivist> if the broken item is loose in the hole but not removable the burr method can grind it out, trying to drill or mill will end up with more scrap
[12:53:39] <piasdom> ya, been dair-done dat
[12:56:23] <holy> How do you call me if I'm scared if while searching the home position, the bit on a Lathe, hits the blank? I mean those awful abrupt movements in -X range of coordinates
[12:57:47] <holy> The laptop is out of battery. Ask, anyway, I will read the log file.
[13:02:08] <nickm_> i
[13:02:16] <nickm_> hi
[13:04:25] <nickm_> I'm lost! :-) I do have a 5i20 (running) and a homebrew ModBus-card with 32 ins and 32 outs. The problem I do have is getting the bits in classic-ladder correctly mapped to the ins and outs on ModBus. There always appear some maskings on the lower or upper half.
[13:05:53] <nickm_> What could be the reason? I'm quite sure that my ModBus works OK. And I get that working with 16 bits IN/OUT rom time to time.
[13:53:55] <clip9> Hm.. is it possible to have the probe pin on the same pin as Home All?
[13:54:33] <BJT-Work> what is a Home All pin?
[13:57:35] <skunkworks_> logger_emc: bookmark
[13:57:35] <skunkworks_> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2009-08-20.txt
[13:58:13] <clip9> It's a pin connected to homing switches on all axis.
[13:59:02] <clip9> One axis can be homed at a time.
[13:59:22] <clip9> I think it's called Home All in Stepconf
[13:59:59] <BJT-Work> I don't think you have the probe and home switches on the same hal pin but I might be wrong
[14:00:15] <BJT-Work> do you have limit switches and home switches
[14:01:49] <clip9> No just the home switches.
[14:03:14] <clip9> I think it's possible. atleast in theory. since the All Home pin is high only when one axis is in the home position. The pin is "free" while the the axis have left the home position.
[14:04:01] <BJT-Work> you also might get an error while homing "probe tripped while not in a probing move" or something to that effect
[14:04:15] <BJT-Work> easy to test out...
[14:04:19] <clip9> yeah
[14:04:23] <clip9> I'll just try it.
[14:07:20] <cradek> I think BJT is right
[14:07:28] <cradek> tripping the probe during homing causes an abort
[14:08:03] <cradek> but with suitable hal or ladder, you could use the home state to disconnect/mask the signal from the probe input
[14:09:00] <jepler> you also have to make sure that the home switches are outside of soft limits, so you never have a switch closed when trying to probe
[14:09:33] <cradek> true - I didn't think of that one.
[14:09:57] <cradek> if your home switches are outside soft limits, make sure the last homing move takes you within the limits
[14:10:39] <clip9> Yep. I've set it up like that already :D
[14:11:08] <clip9> Where should i start reading about setting up hal or ladder?
[14:11:44] <BJT-Work> http://www.linuxcnc.org/docview/html//hal_basic_hal.html
[14:11:49] <jepler> axis.N.homing OUT bit
[14:11:49] <jepler> TRUE if the joint is currently homing
[14:12:18] <BJT-Work> http://www.linuxcnc.org/docview/html//ladder_classic_ladder.html
[14:12:19] <clip9> Thanks..
[14:17:34] <BJT-Work> I guess you could send anxi.N.homing through a not and into an and2 with the probe in... and not need classicladder
[14:18:11] <cradek> yep either way
[14:18:26] <cradek> neither is super simple...
[14:18:38] <BJT-Work> :)
[14:19:44] <clip9> I'll have to try it a bit later. But thanks for the hints :)
[14:20:13] <BJT-Work> * BJT-Work hates windozs updates....
[14:23:23] <geo01005> anybody here know anything about the STEP-NC standard?
[14:26:05] <jymm> Can I use explosives? --> Able to move object up to 1200 lbs
[15:24:57] <alex_joni> geo01005: vaguely
[15:25:07] <alex_joni> I know NIST is working on an interpreter for it
[15:27:57] <geo01005> DanHeeks on the HeeksCAD project noticed the project last night.
[15:28:34] <archivist> http://code.google.com/p/iso-14649-toolkit/
[15:29:26] <geo01005> does anybody know mpictor? I think that I have seen him here on #emc before, but I can't remember.
[15:31:06] <geo01005> He is the main contributor on google code project for STEP-NC that archivist just linked to.
[15:33:25] <skunkworks_> he has emailed me a few time with questions about my h-bridge...
[15:33:26] <skunkworks_> http://www.cnczone.com/forums/showthread.php?p=642792#post642792
[15:33:37] <skunkworks_> if it is the same guy. He has been on here a couple of times.
[15:35:02] <geo01005> he is also the owner of this project: http://code.google.com/p/rs274ngc/
[15:35:17] <geo01005> This is the interpreter that emc uses right?
[15:35:52] <skunkworks_> based on - yes
[15:36:49] <skunkworks_> oh - step as in take a step file and machine it? interesting
[15:37:35] <geo01005> yes, we are looking at an open file format for all the opensource CAD project to use.
[15:38:19] <archivist> heehkcad is young enough to change to that
[15:58:00] <alex_joni> skunkworks_: STEP is a very complete spec
[15:58:13] <alex_joni> you are talking about a small subset of it, that deals with CAD data
[15:58:22] <skunkworks_> ah - ok
[15:58:23] <alex_joni> there is also STEP NC that deals with CAM
[15:58:37] <skunkworks_> alex_joni: puma ran tort.ngc ;)
[15:59:24] <alex_joni> http://www.isd.mel.nist.gov/projects/stepnc/
[15:59:29] <alex_joni> yay ;)
[16:01:09] <cradek> well there went a decent mill... http://www.cnczone.com/classifieds/showproduct.php?product=3207
[16:03:28] <alex_joni> http://www.nist.gov/cgi-bin/view_pub.cgi?pub_id=901160
[16:03:31] <alex_joni> funny date ;)
[16:03:43] <skunkworks_> cradek: :( http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=330348898373
[16:05:07] <skunkworks_> unless - you won them :)
[16:05:18] <cradek> nope, I don't need them
[16:05:27] <cradek> I have a spare nc400 already
[16:05:42] <skunkworks_> oh - whatever - you know you're a horder..
[16:05:54] <cradek> true
[16:05:58] <cradek> only so much space though
[16:05:59] <archivist> whats hording :)
[16:06:07] <cradek> archivist: it's like hoarding
[16:06:14] <alex_joni> gathering?
[16:06:15] <skunkworks_> did I spell it wrong again.. heh
[16:06:24] <archivist> hmm dangerous hobby
[16:06:41] <geo01005> So does this new genserkins modules work for arbitrary serial robots? Or are there exceptions?
[16:06:47] <archivist> houses collapse under the strain
[16:08:40] <skunkworks_> Atleast I got the you're right... ;)
[16:09:03] <alex_joni> geo01005: there shouldn't be any exceptions
[16:09:23] <geo01005> Cool.
[16:09:30] <geo01005> Any docs yet?
[16:09:54] <alex_joni> what? besides the source?
[16:10:10] <geo01005> yes.
[16:10:45] <geo01005> I should go look at the source before I start asked lots of question right :)
[16:12:02] <alex_joni> it's based on this: http://www.roble.info/robotics/serial/html/SerialRobots-1se8.html#x9-190008.1
[16:13:27] <geo01005> Ahh, so it uses a numerical method to solve in inverse kinematics problem.
[16:15:11] <geo01005> Very cool.
[16:15:30] <geo01005> I had hope that this would happen one day.
[16:16:42] <skunkworks_> alex_joni: have you done it to your arm you modeled?
[16:26:07] <skunkworks_> cradek: is the gcode for ball you milled at the fest around?
[16:34:38] <numen> is it possible, to reset to x0y0z0 point of the ngc file in emc2?
[16:35:23] <jepler> numen: g10l2 and g92 both affect the coordinate system as described in the manual
[16:35:44] <BJT-Work> http://www.linuxcnc.org/docview/html//gcode_main.html#sub:G10:-Set-Coordinate
[16:35:47] <jepler> but they don't run the homing procedure or change the machine coordinates
[16:35:57] <BJT-Work> http://www.linuxcnc.org/docview/html//gcode_main.html#sub:G92,-G92.1,-G92.2,
[16:36:19] <numen> cause my gcode file is on y -nn x +nn z +nn
[16:36:26] <numen> and it should by at x0 y0 z0
[16:40:34] <cradek> are you asking whether emc2 will edit your gcode? if so, the answer is no
[16:40:54] <cradek> if you want it to run your program with the coordinates offset by some amount, read the links above
[16:41:59] <jepler> but be aware that "touch off" in the AXIS user interface is just setting the G54 coordinate system using G10L2 MDI commands, so if you G10 L2 P1 ... in your program, "touch off" won't work
[16:42:23] <jepler> if the program has top-of-material as 10.0, then just enter 10.0 instead of 0 when you are touching off Z
[16:47:55] <cradek> a funny misunderstanding about how cooling works: http://thereifixedit.com/2009/08/15/epic-kludge-photo-rube-goldbergs-ac/
[16:50:48] <geo01005> cradek, what do you mean? The fan makes "cold air" and the more cold air you put in the fridge, the colder it gets ;)
[16:52:30] <numen> how is it possible, to add an automatic tool length compensation?`
[16:53:18] <cradek> geo01005: no, they're "cooling" the room by blowing out the freezer's air
[16:53:38] <geo01005> ohhhhh. Ha... yeah that will work well...
[16:54:31] <geo01005> energy balance is a killer :)
[16:55:02] <geo01005> All the while the room is actually getting warmer.
[17:00:52] <elchaschab> hi
[17:01:51] <elchaschab> id like to use emc2 for a "heiz high-z s-720" cnc... but im stuck figuring out the correct parameters for the steppers
[17:02:47] <elchaschab> i already tried to look up the specs for the steppers, but couldnt find all i need... maybe someone help figure out the right parameters?
[17:04:04] <elchaschab> names
[17:06:42] <elchaschab> hi?
[17:07:37] <archivist> you got auto kicked because you probably tried to paste something large elchaschab has quit (SendQ exceeded)
[17:07:57] <archivist> use a pastebin
[17:08:56] <elchaschab> anyone here?
[17:09:05] <archivist> plenty
[17:10:06] <elchaschab> oh... sry... im not used to irc... im currently trying to setup my "heiz high-z s-720" cnc with emc2 but i cant find the right stepper configuration...
[17:10:40] <elchaschab> i already tried to look up the stepper specs, but couldnt find anything suitable... maybe someone could help me to figure out the right values?
[17:10:56] <elchaschab> maybe there is a way to measure or something similar?
[17:11:23] <SWPadnos> not really
[17:11:43] <numen> does anyone know, how i can add an tool offset probe?
[17:11:56] <SWPadnos> it's possible to guess at the torque or power specs based on the size, and you can guess at the voltage by looking at the existing power supply
[17:12:22] <SWPadnos> numen, yes, you can do some tool length probing, but it's not automatic
[17:12:35] <SWPadnos> see the sample file tool_length_probe.ngc in the samples directory
[17:12:44] <SWPadnos> (I think that's what it's called)
[17:13:23] <elchaschab> hmm.. figured that already out... im stuck with the "mechanical" values (microsteps, etc..)
[17:13:38] <SWPadnos> well, you can measure some things
[17:13:52] <SWPadnos> incidentally, it's the motor driver that determines that stuff, not the motor
[17:13:59] <SWPadnos> so you might be able to look those up
[17:14:05] <elchaschab> oh
[17:14:07] <elchaschab> ic
[17:14:24] <elchaschab> so that might be part of the controller specs?
[17:14:39] <SWPadnos> yes
[17:14:42] <SWPadnos> it could be
[17:15:04] <SWPadnos> also, if there are setup parameters in the controller, you can look at those and then translate them to the units EMC2 needs
[17:15:05] <elchaschab> great... thx for the new path ;)
[17:15:08] <SWPadnos> sure
[17:15:18] <SWPadnos> you can determine some of this experimentally as well
[17:15:44] <elchaschab> all ive got is a winpcnc configuration file, that is plain text but kinda obfuscated
[17:15:52] <SWPadnos> if you know the machine is supposed to go up to 20 m/min, then you know that's your goal (to meet or exceed)
[17:15:59] <SWPadnos> ok. that could help anyway
[17:15:59] <elchaschab> hmm
[17:16:38] <SWPadnos> you can determine the steps per inch (or mm) by setting emc to very conservative settings, and say 1000 steps per unit, then measure how far the machine moves when you tell it to go one unit
[17:16:43] <SWPadnos> (which will be 1000 steps)
[17:16:55] <SWPadnos> then divide something or other to get the right number
[17:17:06] <elchaschab> ahh
[17:17:12] <elchaschab> great
[17:17:18] <SWPadnos> once you have that, you can set the velocity to something you're pretty sure the machine can do
[17:17:23] <SWPadnos> err, velocity limit
[17:17:26] <numen> is it not possible, to do this automatic?
[17:17:34] <SWPadnos> then you can experiment with the acceleration setting
[17:17:45] <SWPadnos> no, there's no feedback from steppers, so EMC2 can't tell when they stall
[17:17:53] <SWPadnos> or how far they've gone
[17:18:19] <numen> SWPadnos i mean tool lenght probing
[17:18:23] <elchaschab> ok... thx a lot... only one question left for my... how to determine microsteps per step? or how important is that?
[17:18:34] <SWPadnos> oh sorry - I didn't see who was asking :)
[17:19:00] <numen> i asked for automatic tool length probing, cause i dont have more than 1 tool holder
[17:19:08] <SWPadnos> elchaschab, it's not important by itself. It's only necessary to know so you can get the total number of steps emc2 needs to output for the motor to turn 1 full step
[17:19:11] <SWPadnos> right
[17:19:23] <SWPadnos> numen, look at tool_length_probe.ngc
[17:19:38] <SWPadnos> you can add a call to do whatever it does after each tool change
[17:19:52] <SWPadnos> it's just not automatically going to get run with TxxM6
[17:20:09] <SWPadnos> so you change that to Txx M6 then O<length_probe> call
[17:20:11] <geo01005> Couldn't you write a subroutine?
[17:20:23] <SWPadnos> yes
[17:20:28] <SWPadnos> as above ... :)
[17:20:32] <geo01005> what you just said...
[17:24:28] <numen> so the first tool i have to measure?
[17:24:38] <numen> cant i calculate only the difference?
[17:25:06] <SWPadnos> I don't know the specifics of how it works, you'll have to figure out some of it on your own
[17:25:31] <skunkworks_> if you knew exactly how high your part is off switch - yes - you could do the math automagically.
[17:25:52] <alex_joni> hi samco
[17:25:54] <skunkworks_> I think
[17:25:57] <skunkworks_> Hi alex
[17:26:00] <alex_joni> yeah, I did.. didn't work as fine as the puma though
[17:26:09] <skunkworks_> alex_joni: why?
[17:26:17] <alex_joni> no idea ;)
[17:26:20] <skunkworks_> heh
[17:26:31] <alex_joni> didn't investigate it that much
[17:26:40] <alex_joni> close to joint0 it looks ok
[17:26:53] <alex_joni> but moving away causes the straight lines to bend a bit
[17:27:03] <alex_joni> might be a problem in the kins
[17:27:07] <alex_joni> might be a modelling problem
[17:27:31] <skunkworks_> well - maybe closer to a modelling problem - as the puma seems to work ok...
[17:27:39] <numen> i think, this is not so easy
[17:28:16] <skunkworks_> although it is hard to put a ruler on the screen to see if it was strait.. ;)
[17:29:40] <skunkworks_> numen: I used to do circuit boards that way by changing the coordinate offset (g10). I would touch the switch - cacluate where the tool should be and move the coordinate system. Worked ok but if you can do it by setting the tool length it would be a lot better.
[17:29:44] <skunkworks_> oops
[17:29:48] <skunkworks_> I guess he left
[17:30:05] <archivist> he wants pointy clicky easy
[17:30:32] <BJT-Work> I hate when that happens... your typing away and before you finish he leaves
[17:30:50] <archivist> dunno how we say, engage brain a little
[17:30:53] <SWPadnos> hey archivist, did you say before that you thought a 20.5x1mm tap can be bought?
[17:31:04] <SWPadnos> without having it made special :)
[17:31:17] <archivist> dont think so
[17:31:48] <archivist> but if you can get the object near a machine :)
[17:31:53] <BJT-Work> how about using a thread mill
[17:31:59] <SWPadnos> yeah - that's the other option
[17:32:02] <archivist> exactly
[17:32:23] <SWPadnos> yeah, or a lathe (which is the other option)
[17:32:27] <archivist> Im just thread milling now for the odd stuff
[17:33:48] <archivist> one starts thinking a different way
[17:34:16] <archivist> good excuse to finish retrofitting your mill SWPadnos
[17:34:24] <SWPadnos> not in the next week
[17:34:28] <SWPadnos> (or my lathe :) )
[17:38:28] <archivist> is this just for one nut or what
[17:38:41] <SWPadnos> no, a few aluminum tubes
[17:39:01] <SWPadnos> cradek will do it for me, but I'm looking for options to do it myself
[17:39:06] <archivist> taper pipe thread?
[17:39:10] <SWPadnos> nope
[17:39:26] <SWPadnos> for this: http://www.neutrik.com/us/en/audio/210_629322010/NC3MPR-HD_detail.aspx
[17:39:55] <archivist> good ole xlr
[17:40:01] <SWPadnos> yep
[17:40:45] <archivist> easy enough to cnc a carbon tap for that
[17:40:57] <SWPadnos> ... if your CNC works
[17:41:10] <archivist> :) mine does
[17:41:14] <SWPadnos> but then again, if you have a CNC you can threadmill
[17:44:29] <archivist> play time back in a bit
[17:44:31] <skunkworks_> * skunkworks_ should really configure emc in simulator. (rebooting after jogging the arm past its actual travel ;))
[17:50:07] <geo01005> I can't find this in the docs, at the moment, but can't you store subs in a file rather than adding the same sub to every file you run?
[17:50:26] <SWPadnos> yes
[17:50:36] <SWPadnos> the file should be in yout nc_files directory
[17:50:57] <geo01005> named Oxxx.nc ?
[17:51:04] <SWPadnos> I don't know if it's required, but I think you need to have a subroutine in the file with the same name as the file
[17:51:04] <geo01005> one file for each sub?
[17:51:07] <SWPadnos> sure, or named
[17:51:26] <geo01005> cool.
[17:51:27] <SWPadnos> so if you have a sub called fred, you would have fred.ngc, and you'd call it with O<fred>
[17:51:39] <geo01005> I know I read it once before, but can't find where now.
[17:51:46] <SWPadnos> I don't know if numbered ones need the O in front
[17:51:50] <geo01005> it isn't in the g-code reference.
[17:51:53] <SWPadnos> or if it's just 123.ngc
[17:52:02] <SWPadnos> oh, it should be under O-words or something
[17:54:30] <geo01005> ahh, didn't read far enough... 3.7 Calling Files
[18:37:36] <alex_joni> hi fenn
[18:41:21] <skunkworks_> alex_joni: how do you trap the sigularities?
[18:41:32] <kanzure> hi alex_joni
[18:41:54] <alex_joni> skunkworks_: I don't ;)
[18:41:58] <skunkworks_> heh
[18:42:56] <skunkworks_> How 'would' you..
[18:43:09] <alex_joni> skunkworks_: I'm still pondering (not very much though) how to change emc2 so it does that
[18:43:41] <alex_joni> basicly it (motion?) should try to run the inverse kins at the next position
[18:43:53] <alex_joni> if that fails, it should abort motion, and report an error
[18:44:08] <alex_joni> maybe even switch to joint mode while on a singularity
[18:45:12] <alex_joni> hmm.. this seems like a good idea
[18:45:31] <alex_joni> when switching from joint to world, first try and see if kins report error
[18:45:48] <alex_joni> if they do, abort the switch, switch back to joint mode, and report the error from the kins
[18:46:03] <alex_joni> we probably should have different error codes for various kins problems
[18:46:11] <alex_joni> singularity, convergence errors, etc
[18:47:17] <alex_joni> out of workspace
[18:47:36] <alex_joni> although I'm not sure how that 'looks' in the kins
[18:56:18] <skunkworks_> neat
[19:00:31] <SWPadnos> alex_joni, the "next point" may be just on the other side of a singularity (though I guess that doesn't matter, since motion/kins doesn't deal with the whole path)
[19:14:39] <alex_joni> if it's just on the other side then it's ok :D
[19:16:16] <SWPadnos> as long as the arms don't need to move (nearly) infinitely fast to get there :)
[19:16:54] <alex_joni> going over a singularity should be no problem
[19:17:06] <SWPadnos> yeah - though it depends on the singularity
[19:17:11] <alex_joni> right
[19:17:56] <SWPadnos> for instance, a scara robot has one at the center of rotation of the arm - not only does the math blow up (I think - dividing by a zero radius or some such), but the machine can't instantly reverse the R joint as it crosses over the origin
[19:18:14] <SWPadnos> or spin 180 degrees in a nanosecond
[19:20:32] <alex_joni> one thing about iterative kins is that you don't need to track ambiguities
[19:21:13] <alex_joni> like on a SCARA you could have the arms on the left or on the right
[19:21:37] <SWPadnos> right
[19:21:57] <SWPadnos> though it would be cool if it could "prioritize" motion in the current direction
[19:21:59] <alex_joni> but if you have iterative kins, and start from a set of joint positions .. then it's no issue
[19:22:29] <SWPadnos> sure
[19:23:00] <alex_joni> although I noticed that sometimes.. (for some unknown cause of sometimes) motion passes 0,0,0,0,0,0 as the joint values to the kins
[19:23:17] <alex_joni> for the puma that's right in a singularity, so the kins fail
[19:23:26] <alex_joni> * alex_joni needs to gdb that
[19:23:52] <SWPadnos> think about going to the extent of R motion on a two-arm machine (shoulder+elbow) - as you hit the end, you want the elbow joint to keep moving the way it's moving, so you would actually come back (towards R=0) with the elbow on the other side of the radial line
[19:24:10] <alex_joni> that's another singularity
[19:24:14] <SWPadnos> sure
[19:24:27] <SWPadnos> well, not - it's an ambiguity
[19:24:37] <SWPadnos> s/not/no/
[19:24:38] <alex_joni> complete extended is a singularity
[19:24:48] <SWPadnos> ok - I'll take your word for it
[19:24:58] <SWPadnos> in fact, I should build one of these boards and see if any smoke comes out
[19:25:02] <SWPadnos> see you later :)
[19:25:04] <alex_joni> there are 4 singularities for puma ;)
[19:30:26] <skunkworks> Matt!
[19:30:50] <mshaver> hey!
[19:33:54] <alex_joni> hi matt
[19:34:41] <mshaver> hey Alex!
[19:36:02] <mshaver> I was just looking at my mail and a message I posted to the devel list took at least _several minutes_ to get posted!
[19:38:11] <skunkworks> do you mean - sent to the list? you are lucky. My messages seem to take hours.
[19:38:49] <mshaver> sometimes mine take hours as well
[19:39:30] <mshaver> I had hoped that it would either be really quick, or take hours so I could really complain!
[19:39:37] <archivist_attic> sourceforge needs oiling
[19:42:15] <tom4> tom4 is now known as tom3p
[20:09:09] <alex_joni> mshaver: did you try to check behaviour with other GUIs?
[20:41:15] <numen> re
[20:42:38] <mshaver> alex_joni: do you mean the "position display" behavior, or the "stuck in auto mode" behavior?
[20:47:44] <cradek> mshaver: axis tries to smartly switch modes but it sounds like it's not very successful.. if you need the system in a particular mode for something, you should do that mode switch. what is the original problem you're trying to solve?
[21:02:48] <mshaver> cradek: Actually I'm troubleshooting eztrol. The behavior I described is really no problem in AXIS because you can move from auto mode to manual or mdi mode by clicking on a related widget. In eztrol, the manual tab and all the other tabs in the interface are grayed out when you are in auto mode. The result is that getting stuck in auto mode requires going into estop to get back control of the gui.
[21:07:50] <mshaver> We disabled the rest of the gui during auto mode because some people thought you could pause the running program and then go to manual mode to jog around and other "manual" things. I think the change Wally is going to make will be to enable/disable the tabs based on the interpreter idle state. IF interp is idle then enable the tabs.
[21:10:07] <mshaver> But I wrote this up because it seemed anomalous. I think the mode change at the end-of-program and in response to an abort event is triggered in the task manager, so I thought that an error should cause the same thing to happen.
[21:11:35] <mshaver> In a bit I'll try this with tkemc which I'm pretty sure doesn't manipulate the modes by itself; only if you push F3/F4/F5.
[21:13:15] <numen> anyone here, which uses an automatic toll lenght probe?
[21:13:56] <archivist> numen, on irc ask the real question, not does anyone use xxxx
[21:14:02] <mikegg> not sure if it's documented as a bug... but axis locks on my machine if you try and switch to MDI tab during a homing sequence
[21:14:57] <alex_joni> http://liviualexa.ro/2009/08/de-ce-imi-plac-mie-programatorii/ <- ROFL
[21:16:47] <alex_joni> mikegg: sounds bad.. please file a bug at sourceforge
[21:18:06] <mikegg> oh wait. I think that might have been pre - 2.3.3.
[21:18:26] <mikegg> now it just interupts the homing sequence
[21:18:39] <mikegg> nm...
[21:19:41] <mikegg> heh, sorry! that could have been the start of a great goose chase
[21:21:30] <alex_joni> we usually try to reproduce it, before fixing ;)
[21:23:15] <numen> archivist i want to build me an tool lenght probe, cause i just have 1 tool holder, and if i need more than 1 tool for a part, it does not really work... so i need one, who can explain me, how it works
[21:35:24] <mikegg> numen: http://thewarfields.com/cnccookbook/MTCNCTipsTechniques.htm
[21:35:32] <archivist> numen, google and http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Touch_Probe
[21:35:36] <mikegg> about mid-way down he talks about various touch setting
[21:35:52] <mikegg> just make a simple switch and mount it somewhere on the machine
[21:43:29] <archivist> SWPadnos, other method, get a 20mm tap slit it (3 slots in the groove) small taper pin to expand slightly, "some care required", a method that is applied to reamers
[21:44:23] <numen> i just dont want to set my z axis manual after tool chance
[21:44:32] <numen> change, thats all i want ^^
[21:46:11] <mikegg> i want a hexapod
[21:46:22] <mikegg> go
[22:11:02] <fenn> mikeegg, don't we all
[22:11:17] <fenn> er, mikegg
[22:13:29] <archivist> I dont
[22:15:57] <skunkworks928> me too! Me Too!
[22:16:26] <skunkworks928> hmm maybe I should get some of my other projects done first.
[22:18:03] <skunkworks928> fenn: what where you thinking of using for the hexapod joints?
[22:19:14] <skunkworks928> (as in the end of each actuator)
[22:19:24] <jepler> whee, homing working again on zenbot
[22:19:41] <skunkworks928> jepler: Great! are you happy with it so far?
[22:19:46] <jepler> skunkworks928: yes
[22:19:52] <jepler> I wish it wasn't noisy when moving fast, but I'll live
[22:20:07] <jepler> I still need to get the spindle on-off control working
[22:20:17] <seb_kuzminsky> that's cool jeff
[22:20:27] <jepler> seb_kuzminsky: hostmot2 has worked nicely for this machine as a servo system
[22:20:31] <skunkworks928> did you find your ssr?
[22:20:43] <jepler> I found a spare ssr of the same kind
[22:20:59] <seb_kuzminsky> jepler: yay :-)
[22:27:04] <tom3p> alex_joni: the cd baby rocker is a riot. i never heard of eject -t before! now wheres the emc2 hooks to make it a sliding rack tool changer?
[22:28:06] <archivist> auto change er style collets would be nice
[22:29:44] <mikegg> i could SSH in with my phone to rock the baby. that would be sweet
[22:39:19] <tom3p> the tooltip distance moved by a cartesian xyz machine when commanded G91G1X1Y2Z3F100 is sqrt(dX^2+dY^2+dZ^2) right?
[22:39:19] <tom3p> and the speed being 100, what would the individual velocities be?
[22:39:19] <tom3p> would they be 1/tooltipdist * 100, 2/tooltipdist * 100, 3/tooltipdist respectively?
[22:40:10] <tom3p> 3/tooltipdist * 100 (for Z)
[22:42:09] <jepler> unless it's a trick question, I think that's right.
[22:42:38] <tom3p> no trick, trying to understand how big the voltage might be to each drive
[22:43:05] <tom3p> the velocity command should also be proportional in same way
[22:53:54] <jepler> another way to look at is that if the velocity is v and the distance is d, then the time t = d/v. Since X moves 1 over the course of t, the X velocity is vx = 1/t
[22:54:04] <jepler> neglecting acceleration and deceleration, of course
[23:00:58] <tom3p> i just noticed, if X were the only axis moving the distance, it's vel would be 100
[23:00:58] <tom3p> if X&Y were working together to move the same distance, X & Y both would be slower than 100
[23:00:58] <tom3p> if XYZ etc... as the number of axis achieving the distance increases, the individual velocities decrease
[23:02:05] <fenn_> tom3p: cancel units
[23:02:07] <fenn_> fenn_ is now known as fenn
[23:02:31] <fenn> 100 inches/minute * x inches = 1/minute
[23:03:32] <fenn> the absolute maximum speed is sqrt(n) * max axis velocity
[23:03:37] <skunkworks928> fenn: did you get my earlier question?
[23:03:42] <fenn> no?
[23:04:03] <skunkworks928> I was wondering what you where thinking about using for the joint ends on the hexapod.
[23:04:20] <skunkworks928> were
[23:04:35] <fenn> cardan joints made from either roller skate bearings or polished rod/v-block
[23:04:48] <fenn> well not v-block but a milled vee
[23:05:06] <fenn> the point is that there's no slop
[23:05:34] <fenn> "joint ends" meaning spherical joint right?
[23:06:04] <skunkworks928> yes - so in effect a universal joint
[23:09:57] <tom3p> cool- hexapod laser brain surgery http://www.engr.sjsu.edu/rellis/courses/ME250/ME250pres/hexapod_Shirke.ppt
[23:13:42] <fenn> bleh this took forever to find
[23:14:05] <fenn> here's where i'm drawing inspiration for the hexapod joints http://fennetic.net/irc/pi_m824_blu_i4c_o.jpg
[23:15:57] <fenn> http://fennetic.net/irc/ball-joint.jpg could work for lower precision, higher force
[23:17:04] <fenn> the problem with cardan joints instead of spherical is that the kinematics are slightly different (emc only does spherical joints)
[23:17:44] <fenn> (unless your spider's revolute joints are coplanar) (i think)
[23:18:09] <jymm> is that a word?
[23:18:15] <fenn> what word?
[23:18:20] <jymm> coplanar
[23:18:24] <fenn> yes
[23:18:41] <fenn> means in the same geometrical plane
[23:18:41] <jymm> ok, just asking =)
[23:18:58] <jymm> I believe ya, just wondering is all
[23:19:27] <jymm> (that's not hw I read it initially, thught it was caplonar)
[23:19:50] <jymm> how