#emc-devel | Logs for 2007-12-12

[00:47:58] <jepler> good lord it's bo&dick
[01:13:52] <jmkasunich> lol
[01:14:06] <jmkasunich> not quite
[01:14:50] <jmkasunich> he actually seems reasonably cluefull, just got stuck in a couple places
[01:15:01] <jmkasunich> (that isn't the most intuitive datasheet I've ever seen)
[01:42:57] <jmk-st> ok, so I was wrong
[01:55:24] <cradek> http://timeguy.com/cradek-files/emc/PICT0082.JPG
[01:55:24] <cradek> http://timeguy.com/cradek-files/emc/PICT0083.JPG
[01:55:28] <cradek> ^^ C axis
[02:03:55] <jmk-st> gonna use the time honored "piece of hose" coupling?
[02:09:31] <cradek> not sure. the other one has a solid coupling
[02:09:53] <jmk-st> alignment needs to be darn near perfect for that
[02:10:17] <cradek> well that was a design requirement :-)
[02:10:32] <jmk-st> if there is slop in your motor screws, you could loosen them, couple the shafts and let them align the motor, then tighten the motor screws
[02:10:43] <jmk-st> that only works if both shafts are also perfectly straight
[02:10:44] <cradek> yeah
[02:10:56] <cradek> right, I would tighten the coupling first, then everything else
[02:11:50] <cradek> the shaft is .2499. I have .2497 and .251 reamers
[02:15:06] <jmk-st> try the small one first
[02:15:20] <jmk-st> there are also tricks to make a reamer cut a bit big
[02:15:32] <jmk-st> like a wooden toothpick in the flutes
[02:18:40] <jmk-st> build to print, pound to fit, paint to match
[02:21:36] <SWPadnos> heh
[02:21:49] <SWPadnos> I thought it was "measure with micrometer, mark with chalk, cut with chainsaw"
[02:22:26] <cradek> measure twice, cut once, measure a third time, cut it right the second time
[02:23:00] <jmk-st> assuming you didn't cut it too short the first time
[02:24:01] <SWPadnos> just stretch it
[02:24:25] <jmk-st> cut a little back on
[02:25:53] <cradek> jmk-st: how's yours coming?
[02:26:17] <jmk-st> I'm filing
[02:26:49] <jmk-st> adding the SMI module solved the latency problem, so it back to metalwork
[02:27:02] <cradek> I thought you fixed that once before
[02:27:05] <jmk-st> mounting a molex connector to the plate that mounts the cabinet fan
[02:27:20] <jmk-st> then I'll mount the PC above that plate, and the PC supply can run the fan
[02:27:36] <jmk-st> the SMI? yeah, I did
[02:27:54] <jmk-st> but maybe the 151 updates, or the emc 2.2 update, broke it
[03:21:31] <steves_logging> steves_logging is now known as steve_stallings
[05:25:02] <steve_stallings> steve_stallings is now known as steves_logging
[13:41:12] <Guest191> Guest191 is now known as skunkworks_
[15:04:02] <cradek_> cradek_ is now known as cradek
[15:05:48] <skunkworks_> morning chris
[15:08:15] <cradek> hi
[15:09:28] <cradek> skunkworks_: did you see yesterday's photos of the C axis?
[15:09:55] <cradek> http://timeguy.com/cradek-files/emc/PICT0082.JPG
[15:09:57] <cradek> http://timeguy.com/cradek-files/emc/PICT0083.JPG
[15:17:53] <skunkworks_> cradek: cool - yes I read back. Almost done :) how are you going to make the drive board?
[15:18:23] <cradek> I think I will make one just like the others
[15:18:41] <cradek> I haven't looked to see if I have all the parts yet. (I probably should have done that a few days ago)
[15:21:59] <cradek> it's sure nice now that I've fixed my vise. everything I do comes out square.
[15:22:28] <skunkworks_> nice :)
[15:23:38] <cradek> I want tool length sensing though! Drilling holes is a pain. spot drill, drill .010 under, drill actual size, counterbore. everything is a different length and I don't have enough holders/drill chucks to set lengths.
[15:24:08] <SWPadnos> you need a lathe ...
[15:24:21] <cradek> I do?
[15:24:36] <SWPadnos> err - well, conceptually, you need one. since you have one, that helps a lot
[15:24:57] <cradek> * cradek sends some coffee to SWPadnos
[15:25:05] <SWPadnos> I have coffee, thank you :)
[15:25:12] <SWPadnos> you can make tool holders with a lathe ...
[15:25:14] <cradek> uh-oh, you're still making no sense
[15:25:25] <cradek> oh...
[15:25:53] <cradek> I have two lathes named "too small" and "couldn't turn a correct taper to save its life"
[15:26:18] <SWPadnos> too small is good - use the inaccurate one for roughing, and the small (and presumably reasonably accurate) one for finishing
[15:27:40] <skunkworks_> I was very supprised you guys made the adaptor on your lathe at the cnc workshop
[15:27:51] <skunkworks_> turned out great
[15:28:04] <cradek> skunkworks_: it's small but it's reasonably powerful and accurate
[15:29:51] <SWPadnos> the small screws are a bonus for accuracy
[15:30:02] <SWPadnos> or at least resolution :)
[15:30:14] <cradek> yeah it sure has resolution
[15:30:59] <cradek> I turned a reasonable taper with it to make my tool measuring fixture. I don't get blue throughout, but I do get it at both ends and in the middle
[15:31:31] <cradek> probably the best one can do without some kind of grinding
[15:37:03] <skunkworks_> biab
[17:12:02] <skunkworks_> http://www.cnczone.com/forums/showthread.php?t=48470
[17:12:16] <skunkworks_> http://www.cnczone.com/forums/showthread.php?t=48532
[17:14:37] <jepler> "When you change your steps per unit, you're also changing the rapid speed, even though it doesn't appear to change." -- wtf?
[17:15:00] <skunkworks_> I read that.. scary. (assuming it is right)
[17:17:50] <SWPadnos> hmmm. I know Mach has a few settings for "max step rate" (something like 25000, 35000, 45000 steps/sec), but I don't know if that directly sets the rapid rate
[17:18:05] <SWPadnos> ie, max step rate / steps/unit = rapid speed
[17:19:04] <cradek> that's interesting. if the limiting factor is steps/second, as it often is on stepper machines, that keeps you from having to figure out units differences on your rotaries
[17:19:21] <SWPadnos> ?
[17:19:26] <SWPadnos> you still need to set scaling
[17:19:33] <cradek> yes but how many degrees/second is the same step rate?
[17:19:51] <SWPadnos> * SWPadnos goes for more coffee
[17:20:01] <cradek> it may make sense (but it's very stepper-centric)
[17:20:12] <SWPadnos> mach is quite stepper-centric
[17:20:20] <cradek> sure
[17:20:32] <SWPadnos> it can do closed loop, if you have a galil card (which does closed-loop on its own)
[17:20:51] <cradek> so about gun control
[17:21:01] <SWPadnos> it does this by outputting a series of velocities or positions, much like the EMC motion controller, and lets the Galil card do the work
[17:21:09] <SWPadnos> re: gun control, see the Brady bill
[17:21:13] <SWPadnos> or something
[17:21:20] <cradek> I think gun control is either good or bad
[17:21:31] <SWPadnos> I suppose it could be neutral
[17:21:38] <SWPadnos> but it's much more likely to be good or bad
[17:23:30] <SWPadnos> it's kind of funny actually. Vermont is one of just a couple of states (2 I think - VT and TX) that allows concealed weapons, and we have one of the lowest crime rates in the country
[17:23:45] <jepler> I'm not personally a hypocrite, but I think everyone *should* be one.
[17:23:48] <cradek> NE allows it I think
[17:24:02] <skunkworks_> WI was trying to as is mn
[17:24:07] <SWPadnos> I don't think you need a special license for concealed weapons vs. visible ones
[17:24:27] <SWPadnos> I suspect that most states allow concealed weapons with a special license
[17:25:02] <cradek> my gun-rights-advocate coworker pointed out that the mall in omaha was a 'no concealed weapons' zone (as is much private property)
[17:25:22] <cradek> he wasn't quite bold enough to say that if it hadn't been, fewer people would have been killed because other shoppers would have shot the shooter
[17:25:34] <skunkworks_> actually - I think you can in MN (the thing is every business/goverment building can opt out of it - posting a sign 'no consealed weapons') or something like that. So it makes it pretty worthless.
[17:25:38] <SWPadnos> on a separate note, I finally found some conductive epoxy, so I can glue the rear defroster tab back on the glass instead of replacing the entire rear windshield
[17:26:12] <cradek> did anyone else notice that the words "shopper" and "shooter" look the same?
[17:26:20] <skunkworks_> nice - way to change the subject ;)
[17:26:25] <skunkworks_> heh
[17:26:31] <SWPadnos> which is good, because the glue/repair kit cost $20, and a used widnshield is $400, or a new one is $700
[17:26:33] <cradek> I'm immune to your subject-changing
[17:27:00] <cradek> SWPadnos: mine has a couple broken 'lines'... I wonder if I could use the same stuff
[17:27:14] <SWPadnos> yes, the kit has conductive goo for the resistive strips as well
[17:27:32] <SWPadnos> available at any auto parts store, in the section where they have the rear-view mirror glue
[17:27:45] <SWPadnos> it's Permatex part number 09117
[17:28:10] <SWPadnos> "Complete Rear Window Defogger Repair Kit"
[17:28:21] <cradek> thanks, maybe I'll try
[17:28:38] <cradek> unfortunately tinting is involved on my back window. It may not be accessible.
[17:28:42] <cradek> a?
[17:29:22] <skunkworks_> cut a little bit out (if it is film)
[17:29:39] <skunkworks_> cradek: been updating this thread here http://www.cnczone.com/forums/showthread.php?t=47646
[17:29:43] <cradek> skunkworks_: yeah, if I can find the break
[17:30:10] <skunkworks_> ah - i figured you kew
[17:30:12] <skunkworks_> knew
[17:31:12] <cradek> hmm maybe I should replace those 2MB photos then...
[17:33:17] <skunkworks_> heh - I was going to scale it... but lost interest ;)
[17:34:56] <skunkworks_> if you made it smaller - I would embbed it in the reply
[17:40:20] <cradek> o100 repeat [10]
[17:40:20] <cradek> g91g0x.1
[17:40:20] <cradek> o100 endrepeat
[17:40:30] <cradek> does anyone but me think this would be nice?
[17:40:57] <SWPadnos> yes, if there's no "for" loop already
[17:40:57] <fenn> why not a for loop?
[17:41:14] <cradek> this is almost always how I use while[], but it's very tedious to do a counter and comparison
[17:41:22] <cradek> right, there is no for loop
[17:41:28] <SWPadnos> yes, the "unnamed variable" would be nice
[17:41:43] <cradek> SWPadnos: I don't follow
[17:42:41] <SWPadnos> ie, repeat[10] creates an unnamed variable, which you don't explicitly need to decrement and compare
[17:42:50] <cradek> ok, right
[17:43:42] <cradek> of course it's evaluated so you can do repeat[sqrt[2]]
[17:43:48] <cradek> (rounds to int)
[17:44:01] <cradek> I already have it working, fwiw, I'm trying to decide whether to check it in
[17:44:14] <jepler> cradek: I am surprised you'd even write it
[17:44:18] <fenn> sure why not
[17:45:04] <jepler> I have a feeling that the "loop" construct that would be nice is one that mirrors the calculation of successive depths in a pecking configuration. It's not easy to get right.
[17:45:18] <jepler> e.g., cut down by 0.1" each time to 2.375"
[17:45:41] <cradek> yeah, ick
[17:45:51] <cradek> good thing we have G83 (which is thankfully written in C)
[17:46:24] <fenn> is there a canned cycle that leaves the bit at the bottom of the hole?
[17:46:24] <jepler> yeah but you want the same kind of calculation in a pocketing routine that can't do the whole depth at once
[17:46:48] <SWPadnos> a while loop seems fine for that kind of thing though
[17:46:58] <SWPadnos> since you probably need to clip the last move a little
[17:47:20] <SWPadnos> also, you need access to the loop variable within the loop, since it's the new depth
[17:47:29] <jepler> yes, you do
[17:49:12] <cradek> http://timeguy.com/cradek-files/emc/nested.png
[17:49:21] <cradek> ok now I've tested this 100%
[17:49:37] <jepler> haha
[17:49:44] <SWPadnos> does it properly catch errors like improper nesting?
[17:49:52] <SWPadnos> ie, swap lines 6 and 7
[17:49:59] <cradek> dang you
[17:50:01] <SWPadnos> heh
[17:50:33] <SWPadnos> same 10(?) levels as other calls/loops?
[17:50:38] <cradek> uh, it doesn't error, it does something strange instead
[17:50:49] <SWPadnos> kewl
[17:51:21] <cradek> http://timeguy.com/cradek-files/emc/gigo.png
[17:52:20] <jepler> 'cut down by 0.1" each time to 2.375"' will take 24 passes. I assume the current peck drilling cycle would do 23 with a depth increment of 0.1 and one with a depth increment of .075. Would it be acceptable if instead they were all an equal 0.9896?
[17:53:23] <fenn> no, because the user asked for 0.1
[17:53:32] <SWPadnos> that's probably OK but it seems unnecessary to do that calculation
[17:54:05] <fenn> maybe it works for peck drilling, but say you are doing threadmilling or something..
[17:54:15] <SWPadnos> the other question is whether it's better to do the oddball pass first, and then step down by the user-requested amount to the user-requested depth
[17:55:09] <cradek> if anything it's important that the first pass be the requested value
[17:55:24] <cradek> the first peck is the one where the drill centers itself etc. you don't want to screw it up
[17:55:29] <SWPadnos> sure
[17:55:47] <SWPadnos> I guess pecking 0.001 might not be ideal, but then again it shouldn't hurt anything either
[17:56:02] <cradek> it could screw up your spot drill that you just did
[17:56:37] <jepler> that was another question I had -- if the total depth is 2.40001, should there be a .00001 pass after the last .1 pass?
[17:56:52] <cradek> heh
[17:57:02] <cradek> define 'should'
[17:57:06] <SWPadnos> not if it's < DEADBAND ;)
[17:57:34] <jepler> (which is actually very relevant because sum([0.1]*10) < 1.0))
[17:57:36] <cradek> you mean if it's less than one microstep
[17:59:57] <SWPadnos> a NANOstep
[18:02:39] <cradek> huh, if I write mismatched endwhiles, I also don't get an error
[18:02:45] <jepler> here come the geese!
[18:03:34] <jepler> (about two dozen just flew in, probably to hang out in the pond we have here next to the building)
[18:16:11] <skunkworks_> isn't the pond frozen?
[18:30:45] <LawrenceG> cradek: I would like to see repeat as an o word primitive... check it in unless you feel inspired enough to do a real "for loop"
[18:57:47] <skunkworks__> skunkworks__ is now known as skunkworks_
[20:02:00] <CIA-42> EMC: 03cradek 07TRUNK * 10emc2/src/emc/rs274ngc/ (interp_read.cc interp_o_word.cc interp_internal.hh): repeat/endrepeat O words to make simple loops simpler than using while
[20:04:29] <cradek> seb_kuzminsky: please use lowercase directory and file names in cvs
[20:05:01] <seb_kuzminsky> okay - i'm used to bzr... sorry
[20:05:57] <cradek> you could make a new directory 'source' and I will remove 'SOURCE'
[20:07:02] <seb_kuzminsky> heh
[21:23:14] <CIA-42> EMC: 03seb 07TRUNK * 10emc2/src/hal/drivers/mesa7i43-firmware/ (7 files): initial commit of upstream files from Mesa Electronics
[21:24:20] <CIA-42> EMC: 03seb 07TRUNK * 10emc2/src/hal/drivers/mesa7i43-firmware/SOURCE/ (6 files): initial checkin of upstream source for the eppio fpga firmware, from Mesa Electronics
[21:24:45] <SWPadnos> seb_kuzminsky, I thought that was supposed to be in a lowercase directory name ;)
[21:24:51] <cradek> arg
[21:25:49] <cradek> I can fix it
[21:25:56] <alex_joni> hi guys
[21:26:19] <alex_joni> is that the stuff Luke uses?
[21:26:35] <CIA-42> EMC: 03seb 07TRUNK * 10emc2/src/hal/drivers/ (mesa7i43-gpio.h mesa7i43_gpio.comp Submakefile): new driver for the Mesa 7i43 FPGA board with EPP interface
[21:27:14] <seb_kuzminsky> swpadnos: i thought so too - i've got a "SOURCE" and a "source" dir locally, and i put the stuff in "source" and added it there
[21:27:22] <seb_kuzminsky> but I guess CVS think it knows better ;-)
[21:27:51] <seb_kuzminsky> anyone wanna have a revision control system flamewar? :-)
[21:27:57] <cradek> no
[21:28:03] <alex_joni> no thanks :)
[21:28:24] <cradek> let me fix it in the repository
[21:28:33] <seb_kuzminsky> thanks I'd appreciate that
[21:28:33] <cradek> then you will have to mess with your checkout until it's happy again
[21:28:40] <seb_kuzminsky> sorry for the hassle
[21:30:23] <cradek> fixed
[21:30:25] <cradek> no problem
[21:30:31] <cradek> when I did an update, I got a source directory
[21:30:41] <cradek> I'm not sure what you will have to do
[21:31:18] <seb_kuzminsky> i've used cvs extensively in the past, i think i can get it working here
[21:31:43] <cradek> if nothing else, deleting your entire mesa7i43-firmware directory and cvs up -dP will fix it
[21:31:55] <seb_kuzminsky> thanks for fixing the repo...
[21:31:59] <cradek> welcome
[21:33:27] <seb_kuzminsky> ok i'm all up to date over here. i'll try not to hose it again...
[21:33:59] <cradek> no worries
[21:34:07] <alex_joni> seb_kuzminsky: don't forget to add yourself to AUTHORS
[21:34:19] <alex_joni> seb_kuzminsky: and to add your changes to emc2/debian/changelog
[21:34:51] <alex_joni> seb_kuzminsky: hi btw ;) didn't get a chance to talk so far
[21:51:21] <seb_kuzminsky> hi alex, i'll do that
[22:10:36] <CIA-42> EMC: 03seb 07TRUNK * 10emc2/src/hal/drivers/rbf2h.py:
[22:10:36] <CIA-42> EMC: Make LGPL work, and a minor cosmetic fix for GPL.
[22:10:36] <CIA-42> EMC: (I discovered these when shamelessly stealing rbf2h.py for mesa7i43's
[22:10:36] <CIA-42> EMC: to-header.py.)
[22:16:50] <fenn> * fenn squints at the actual code-writing part of rbf2h.py
[22:17:41] <fenn> that's pretty neat
[22:19:07] <seb_kuzminsky> about the debian/changelog: looks like multimaintainer mode is off, is that intentional?
[22:27:56] <alex_joni> well.. I'm off to bed
[22:27:59] <alex_joni> good night all
[22:30:32] <seb_kuzminsky> see ya alex
[22:31:46] <CIA-42> EMC: 03seb 07TRUNK * 10emc2/debian/changelog: Added a note to the changelog about the new 7i43 driver, added myself to AUTHORS
[22:31:46] <CIA-42> EMC: 03seb 07TRUNK * 10emc2/docs/AUTHORS: Added a note to the changelog about the new 7i43 driver, added myself to AUTHORS
[22:57:57] <jepler> seb_kuzminsky: the mode where it puts [Your Name] above the change you are adding to debian/changelog? In my view, CVS preserves information about who made what change; the changelog is for public consumption and generally it is unimportant which developer made a particular bug fix or new feature -- all the user reading the changelog wants to know is what the differences *are*
[23:14:09] <seb_kuzminsky> jepler: that seems reasonable
[23:59:20] <seb_kuzminsky> bye all