#emc-devel | Logs for 2007-08-29

[00:02:42] <SWPadnos> is that case 1, 2, or 3? (from thatlast line in my big blathering :) )
[00:03:25] <skunkworks> umm - have not gotten that far yet :)
[00:03:31] <SWPadnos> okie dokie ;)
[00:08:53] <skunkworks> http://cnczone.com/forums/showthread.php?t=37309
[00:11:03] <skunkworks> SWPadnos: the 3d_chips run was 3
[00:11:09] <SWPadnos> ok
[00:11:49] <skunkworks> Looks like I have to slow z acc down a bit.. It is a bit unreliable. which is a bit odd because it is 1:2 - yes geared up.
[00:12:48] <skunkworks> I would have thought it would have been in its lower torque range
[00:13:21] <SWPadnos> sure, but the torque load is still higherwith 1:2 gearing than it would be with 1:1
[00:13:33] <skunkworks> I guess
[00:14:34] <SWPadnos> seems funny to gear up Z - I'd think an engraver wouldn't do too much Z motion, and that the shorter travel range would make it doubly unnecessary
[00:14:47] <skunkworks> no clue.
[00:20:16] <skunkworks> http://www.electronicsam.com/video/PICT1181.AVI
[00:20:24] <skunkworks> real 7.5 in/s
[00:21:20] <skunkworks> per axis
[00:21:27] <SWPadnos> cool
[00:22:09] <SWPadnos> let's see. if I used geckos, the embedded core 2 duo PC, and doublestep, I could get ...
[00:22:12] <SWPadnos> (think think think)
[00:22:30] <SWPadnos> 150 IPM per axis
[00:22:34] <SWPadnos> roughly
[00:22:40] <skunkworks> thats all?
[00:22:46] <SWPadnos> 40000 steps/inch
[00:22:49] <skunkworks> ah
[00:23:11] <skunkworks> that is with 7.5 - stepgen 8
[00:23:31] <SWPadnos> I was assuming 10000 BASE_PERIOD, though it's possible I could increase that
[00:23:41] <SWPadnos> err - decrease - you know, make it faster
[00:23:47] <skunkworks> heh
[00:23:58] <skunkworks> this is 30000.
[00:24:09] <skunkworks> I think this computer is a 1.7ghz.
[00:24:16] <skunkworks> I should try it on the 1ghz
[00:24:41] <SWPadnos> the C2 duo is dual core 1.83 GHz, and I probably can only get the very high rate if I run it headless
[00:24:52] <SWPadnos> or at least X-less
[00:25:19] <skunkworks> ah
[00:25:22] <SWPadnos> though interestingly, it didn't seem to be X that caused problems (in the little testing I did with vs. without)
[00:25:41] <skunkworks> can you isolate one of them?
[00:25:44] <SWPadnos> it was all the other crap they load for a "fully functional desktop environment" that did it
[00:25:49] <SWPadnos> yes
[00:26:03] <SWPadnos> when I did that, and loaded down the other core, the latencies were usually below 200 ns
[00:26:10] <skunkworks> wow
[00:26:33] <SWPadnos> with periodic jumps to ~5-6 us, but I think that's kjournald, which I can disable by using ext2 instead of ext3
[00:28:42] <skunkworks> 7.5 ips is 1800 rpm at the steppers
[00:28:58] <SWPadnos> full step?
[00:29:09] <SWPadnos> (no, doublestep :) )
[00:29:28] <skunkworks> 1000 steps per rev
[00:29:32] <SWPadnos> ok
[00:29:43] <SWPadnos> so you're at 1/10 the resolution of my mill
[00:29:44] <skunkworks> input scale 4000 (4 treads per inch)
[00:29:52] <skunkworks> it is pretty cource
[00:29:59] <skunkworks> course
[00:30:00] <skunkworks> whatever
[00:30:08] <SWPadnos> coarse
[00:30:15] <skunkworks> these poor things - I am very suprisedhttp://www.electronicsam.com/images/engraver/stepper.JPG
[00:30:19] <skunkworks> http://www.electronicsam.com/images/engraver/stepper.JPG
[00:30:20] <SWPadnos> chunky ;)
[00:30:47] <skunkworks> 6 wire - only running 2 of the 4 coils.
[00:30:57] <skunkworks> have not tried series yet
[01:37:59] <SWPadnos> cradek, do you suppose it's time to gently prod Michael for results?
[02:14:11] <skunkworks> I had to decrease the z accelleration to 35 - now it seems to be reliable.
[02:14:33] <skunkworks> over the distance it seems to only make it up to say 270ipm or so
[02:15:04] <skunkworks> 1.7 inches ;)
[02:15:31] <cradek> I noticed it has really short Z travel
[02:15:42] <skunkworks> total is like 2.25 in the specs
[02:16:40] <skunkworks> did you see the video - that is each axis set to 7.5ipm - doublefreq is cool ;)
[02:17:23] <skunkworks> and it does reach 450 ipm and the combined speed is correct
[02:17:50] <cradek> great
[02:18:34] <cradek> that 95% thing was stupid - the better similar fix would be to put 105% in stepgen
[02:18:43] <cradek> or whatever number is needed
[02:18:48] <cradek> * cradek grumbles about this problem
[02:19:17] <skunkworks> it doesn't seem to have a good solution - well that would make everyone happy :)
[02:19:33] <cradek> very true
[02:19:42] <SWPadnos> I think so far it looks as though running with stepgen.maxvel = 0 works
[02:19:47] <SWPadnos> but accel has to be set
[02:19:52] <cradek> that doesn't make me stop being unhappy about it though
[02:20:09] <skunkworks> yes - although with the little testing - it stalled on me more likely than having a headroom.
[02:20:10] <SWPadnos> it's possible that setting accel to some exceedingly large number (like 5000) will also work
[02:20:30] <skunkworks> but I have not tested it compleatly yet :)
[02:20:30] <SWPadnos> was that with accel limited and vel not limited, or with accel not limited as well?
[02:20:38] <SWPadnos> I thought it was accel=0 that did that
[02:20:52] <skunkworks> I only tested both 0.
[02:21:00] <skunkworks> so far
[02:21:14] <SWPadnos> ok. vel=0 accel=5000 would be the next one to check (when you get a chance)
[02:21:24] <skunkworks> ok
[02:57:04] <skunkworks> SWPadnos stepgen vel = 0 seems to work fine.
[02:57:25] <SWPadnos> excellent
[02:57:42] <SWPadnos> with accel set "normally" or to some very high number?
[02:58:11] <skunkworks> stepgen accellleration set to either 0 or a large number 5000 seems to not be as smooth - and I get stalls. I have run it with stepgenaccellertaion set to about 10 over and can run it over and over again without a stal
[02:58:36] <skunkworks> I have had vel set to 0 for all these tests
[02:58:41] <SWPadnos> interesting. that's probably the real accel value for the machine then, at least unloaded
[02:59:12] <SWPadnos> ok. so it sounds like the accel limit can't be "disabled", but the limit may not be as important as we had thought
[02:59:53] <skunkworks> the thing is - it stalls durring the cruise phase it seems.. so does it bounce around more?
[03:00:10] <SWPadnos> hmmm
[03:00:13] <skunkworks> heh
[03:00:14] <SWPadnos> it may
[03:00:32] <skunkworks> but It all happens so fast... I may not be sure.
[03:00:33] <SWPadnos> I'd scope - err - something to find out
[03:00:35] <skunkworks> :)O
[03:00:56] <skunkworks> did you see the video?;)
[03:00:57] <SWPadnos> stepgen has a frequency output doesn't it?
[03:00:59] <SWPadnos> yep
[03:01:02] <SWPadnos> very nice
[03:01:37] <skunkworks> like I say - it doesn't sound as smooth
[03:01:38] <cradek> sounds like we could safely set stepgen accel to the axis limit + 10% or so
[03:01:49] <cradek> and stepgen maxvel to 0
[03:02:07] <cradek> but jmk may not like it. we'll have to see what he says.
[03:02:11] <SWPadnos> I think so
[03:02:12] <SWPadnos> heh
[03:02:43] <skunkworks> I don't know how close - I am running about 120% 50-60
[03:04:25] <skunkworks> something like - if stepgen accelleration = 0 then axis_maxaccelleration *1.2 or something.
[03:04:54] <SWPadnos> I think stepgen.maxvel defaults to 0
[03:05:09] <skunkworks> where the heck is jmk - it is like he has a life or something.
[03:05:20] <cradek> good question - hope everything is ok.
[03:05:22] <skunkworks> present company excluded of cource
[03:05:25] <SWPadnos> this would be a great place for a python version of halcmd - it would be easier to add math to the commands
[03:05:33] <SWPadnos> he's traveling today
[03:05:57] <SWPadnos> the jmk-wvm was likely his laptop from a hotel room
[03:06:55] <cradek> SWPadnos: surely you mean lisp
[03:07:11] <SWPadnos> well, I don't like curly braces much
[03:07:17] <skunkworks> so - a nice video of it cutting a circuit board and off to ebay it goes I think..
[03:07:27] <SWPadnos> use a tripod :)
[03:07:31] <cradek> skunkworks: what's the spindle speed?
[03:07:36] <skunkworks> 20k max
[03:07:47] <skunkworks> it has a bit of a speed control on it.
[03:07:48] <cradek> darn, you can't cut a pcb very fast then
[03:08:05] <cradek> hook the spindle to 220 for a while
[03:08:08] <skunkworks> heh
[03:08:32] <cradek> just long enough for a video of course
[03:10:08] <skunkworks> the specs say 18K :(
[03:10:37] <skunkworks> asychronous 125w
[03:11:20] <skunkworks> the motor is probably 3600 rpm or something - it gets geared up by atleast 2 pully
[03:11:23] <skunkworks> pullys
[03:17:19] <skunkworks> so - did you say .0005 per tooth - that is .001 per rev - that calculates out to 18ipm I think
[03:17:34] <cradek> yep
[03:17:52] <cradek> much faster and the tiny tip breaks off
[03:18:02] <skunkworks> heh -
[03:18:14] <skunkworks> how fast is the spindle on your max?
[03:18:15] <cradek> I know about this from when my spindle was more like 8k
[03:18:19] <cradek> it's near 20k now I think
[03:18:22] <skunkworks> ah
[03:18:36] <skunkworks> so do you run about that for speed?
[03:18:41] <cradek> now ask me what the rating on the bearings are
[03:19:24] <cradek> (roughing feed) #11=15
[03:19:24] <cradek> (outline feed) #12=20
[03:19:24] <cradek> (drill feed) #13=8
[03:19:24] <cradek> (dimension feed) #14=12
[03:19:24] <cradek> (mill-drill feed) #15=8
[03:19:39] <cradek> so yes, 20 I guess
[03:19:41] <skunkworks> i should have known that
[03:20:19] <skunkworks> oh well - time for bed. Thanks again for all your hard work.
[03:20:25] <cradek> :-)
[03:20:32] <cradek> goodnight
[03:20:49] <skunkworks> night
[11:40:50] <cradek_> cradek_ is now known as cradek
[12:07:20] <skunkworks> heh - 2 more votes.. All I have to do is bump one of you guy off for the next election.
[12:07:49] <skunkworks> oops 3 I mean
[13:32:45] <jepler> skunkworks: or talk someone into retiring
[13:32:49] <jepler> I'm sure cradek will be tired of emc2 in another year
[13:56:38] <skunkworks> heh - I hope not. He still has the bridgeport to convert.
[13:57:12] <cradek> I sure hope that's working in a year - who knows.
[13:57:25] <cradek> if it didn't work *at all* now, I bet it would get done faster
[13:58:00] <skunkworks> well - we still have a non working mill ;)
[13:58:41] <skunkworks> is there going to be any more 2.1.x releases?
[13:59:29] <cradek> I think there is an unreleased bugfix in that branch
[13:59:40] <cradek> (I'd have to look to remember what it is)
[14:00:13] <SWPadnos> comp error reporting
[14:00:40] <cradek> I know that doesn't answer your question but it's the best I've got :-)
[14:01:12] <SWPadnos> some image-to-gcode documentation thing was fixed
[14:01:16] <skunkworks> heh - I would assume doublefreq will only be put into the 2.2 release.
[14:01:37] <SWPadnos> well, there's got to be *some* reason to upgrade ;)
[14:01:47] <SWPadnos> gotta keep that revenue stream flowing
[14:02:22] <jepler> there's also merge rev 1.7: properly restore parameters at subroutine end
[14:03:08] <jepler> cradek: I want a pony -- can the branch be added to the emc-commit subject line?
[14:03:40] <jepler> this one should probably be merged if we're making another 2.1: Forbid MDI commands in OFF, ESTOP, and ESTOP RESET states (fixes http://sourceforge.net/tracker/index.php?func=detail&aid=1743942&group_id=6744&atid=106744)
[14:03:52] <jepler> http://cvs.linuxcnc.org/cvs/emc2/src/emc/task/emctaskmain.cc.diff?r1=1.95;r2=1.96
[14:10:13] <cradek> it couldn't hurt to merge it either way.
[14:17:30] <cradek> while read i; do case $i in $USER:*) NAME=${i#*:*:*:*:}; NAME=${NAME%%:*};; esac; done </etc/passwd
[14:17:34] <cradek> * cradek cries
[14:18:36] <cradek> jepler: I added the branch to the subject
[14:18:40] <cradek> I think.
[14:23:07] <jepler> cradek: not sure if this is better or not .. http://pastebin.ca/674743
[14:27:51] <jepler> cradek: thanks for that
[14:27:51] <SWPadnos> what is the intent of that bash line? (I think I can mostly see what it does, but I don't know why one would do it)
[14:28:13] <cradek> extract a real name from a login name using no external programs
[14:28:24] <SWPadnos> ah, ok
[14:28:31] <cradek> and not screwing up $*
[14:28:32] <SWPadnos> I guess I didn't really see what it was doing :)
[14:29:20] <jepler> because mine's in a shell function it doesn't disturb the global $*
[14:29:56] <cradek> right, yours is probably better, but mine is already working
[14:30:19] <jepler> yeah indeed
[14:30:26] <jepler> don't disturb a shell script that works
[14:31:20] <cradek> http://pastebin.ca/674754
[14:31:34] <cradek> where I could really use your help is here...
[14:31:52] <cradek> this is tested and works (copied into another script and run on a local rcs file)
[14:32:03] <cradek> but on the cvs server it never prints the "suppressed" message
[14:32:25] <cradek> and I have never been able to figure out the problem
[14:33:43] <cradek> the trim at 100 lines works, so the builtin [] does have -le, so I assume it has -gt as well
[14:34:59] <jepler> you could try -le || instead of -gt && if you thought there was any chance -gt wasn't available
[14:35:19] <cradek> true
[14:35:23] <jepler> but I agree that seems .. unlikely
[14:35:27] <cradek> right.
[14:36:46] <SWPadnos> being a shell novice, I'd try making the case EEOODD*)
[14:36:57] <SWPadnos> just in case :)
[14:37:00] <cradek> *EEOODD* even
[14:37:21] <SWPadnos> do you have the same shell running on both machines?
[14:37:38] <cradek> I think so
[14:37:45] <cradek> different builds, but same bsd sh
[14:38:12] <SWPadnos> there could also be some newline problem causing the EEOODD to not get read correctly or something silly like that
[14:38:26] <cradek> it's true I mess with IFS when reading
[14:39:57] <SWPadnos> hmm. if diff doesn't output a newline at the end, the EEOODD won't be at the beginning of the line
[14:40:17] <SWPadnos> so a minor version mismatch in diff/cvs could also be problem
[14:40:32] <jepler> fwiw I don't spot any problems
[14:40:43] <SWPadnos> heh
[14:41:00] <cradek> I added the **
[14:41:12] <SWPadnos> as they say - when debugging, put on your stupid hat. for me, that's the only hat I have related to unix shells :)
[14:42:46] <cradek> I wonder why I wanted that inside the case. It could be afterward I think.
[14:43:17] <jepler> scope of variable in ... | while read ?
[14:43:52] <cradek> yep
[14:44:50] <SWPadnos> but linecount is declared outside the loop ...
[14:58:48] <jepler> hahah declared
[15:01:57] <SWPadnos> err - initialized?
[15:45:22] <jepler> in 'x | y', y is in a subshell, and changes to its environment don't propagate back up.
[15:45:25] <jepler> sh -c 'X=1; echo | X=2; echo $X'
[15:45:32] <jepler> in fact both x and y are in subshells
[19:33:07] <SWPadnos> hiya jmkasunich
[19:33:13] <SWPadnos> (if you're actually there :) )
[19:34:10] <cradek> eh he's not there.
[19:34:14] <SWPadnos> heh
[19:34:28] <skunkworks> oh - he has been in and out for the last few weeks. I feel the same way.
[19:35:00] <SWPadnos> I know he was traveling yesterday, but I think this is an IRC client thing - it gets disconnected from time to time and seems to be set to reconnect automatically
[23:51:56] <tfmacz> LawrenceG: You there???