#emc-devel | Logs for 2009-12-14

[00:24:40] <cradek> micges: did you verify that they work?
[00:24:44] <cradek> er, he's gone
[01:17:05] <elson> Hello, all!
[01:17:44] <elson> I was going to try out Anders Wallin's inv. deadband, but I can't get comp to work.
[01:18:45] <elson> On an old system, I get errors that it can't find "component", but that is the first word in the file, on the system I revently updated, even after doing ". emc-environemnt", it doesn't know the command "comp"
[01:19:44] <elson> hmmm, anybody home?
[01:21:28] <skunkworks> jon!
[01:22:27] <elson> Hello, shunk!
[01:22:37] <elson> Oops, Skunk!
[01:23:00] <skunkworks> I have some B series amc drives that I am thinking of getting the brushless servos you have linked to.. For playing
[01:24:29] <elson> You mean the Keling motors? They are pretty affordable, and the Cui encoders go on pretty easily.
[01:24:42] <skunkworks> yes
[01:25:31] <elson> Well, that's what I have hooked up on my bench right now, I wanted to look at this deadband issue.
[01:28:03] <skunkworks> I might get a few for a circuit board mill project http://www.electronicsam.com/images/KandT/pcbmill/tableanddisp.JPG
[01:28:11] <skunkworks> some assembly required
[01:29:43] <elson> A cable driven Something!
[01:31:28] <elson> So, where is everybody?
[01:32:46] <mozmck> elson, do you have emc-dev installed?
[01:35:13] <elson> Yes, I installed emc-dev and build-essential
[01:35:53] <cradek> skunkworks: cool stuff
[01:35:54] <mozmck> hmm, I don't know what the problem is then.
[01:36:19] <mozmck> skunk, I recognize that controller!!
[01:36:28] <cradek> chris@rover2:~$ . ~/emc2.trunk/scripts/emc-environment
[01:36:28] <cradek> chris@rover2:~$ which comp
[01:36:28] <cradek> /home/chris/emc2.trunk/bin/comp
[01:37:05] <cradek> in other words, comp is built along everything else. check that your build was successful.
[01:37:10] <mozmck> I just got a working Asymtek 402G that looks nearly just like that at an auction for $150
[01:37:49] <elson> yes, I did the emc-environment command on both systems. the comp in question is Anders Wallin's idb.
[01:38:36] <cradek> what's your exact error message?
[01:38:49] <mozmck> skunkworks, you wouldn't have happened to have gotten the fluidmove software with that would you?
[01:38:50] <elson> Well, on the newer system I compiled the development head from git source, and EMC runs fine.
[01:39:14] <cradek> do you mean you're getting "bash: comp: command not found" or something else?
[01:39:59] <elson> On the newer system, I get sudo: comp: command not found
[01:41:09] <elson> on the older system, which is fairly old, comp seems to try massaging the source file, but compains it can't find any of the keywords, like "component", pin, etc.
[01:41:44] <cradek> I think comp files have sometimes changed format at major version changes
[01:42:39] <elson> Yes, Anders' file looked a little different syntac, but I made it look more like the examples in the hal guide, and it still complains it can't find "component".
[01:42:41] <cradek> on the new system, if you're using run-in-place (which I deduce from your sourcing emc-environment) I doubt you should be using sudo
[01:43:14] <elson> Ahhh, that could be it, steps out of the environment! Just a minute, let me try it...
[01:43:23] <cradek> sudo may sanitize the environment, removing your oddball path, for sfaety
[01:43:32] <cradek> and/or safety
[01:44:37] <cradek> skunkworks: what are the guides that slide on those rods? are they bearings or just something like delrin?
[01:46:28] <cradek> elson: I am working on mounting a regular encoder to the spindle. If I succeed you can borrow or have both parts of that resolver. The outside part is permanently in the motor cap, but I'll make a new cap.
[01:47:16] <elson> Thanks, Chris, not doing sudo fixed that problem, comp now works. I may have to fix up some syntax in Anders' file to make it compile.
[01:48:14] <elson> The older system may just need an EMC update, but I'm in a hurry to try some stuff, have to go out of town for work tomorrow.
[01:48:34] <cradek> hope the weather is good to you
[01:49:47] <elson> As for the resolver thing, no hurry, I just sent out for 50 boards. I've made some slight changes that might allow it to deliver a little more drive to the resolver, but I think it won't be enough for those VR units. But, I'd like to try it and find out how much current they really take.
[01:50:38] <cradek> ok. hope you got the board changes right for 50! wow.
[01:51:50] <mozmck> cradek: the asymtek I have says it will do about 1200 in/min - would delrin work for that?
[01:52:25] <cradek> mozmck: probably, but I'm no expert
[01:52:39] <cradek> that's why I wondered what skunkworks's was
[01:53:05] <cradek> I have a bunch of delrin and I'm always tempted to make a little gantry with a high speed spindle for engraving
[01:53:28] <cradek> seems like carefully bored holes in delrin + TGP rod would be a setup that's pretty easy to build
[01:53:43] <mozmck> I see. I plan to use my machine to dispense solder paste on small runs of circuit boards.
[01:53:55] <cradek> neat
[01:54:29] <mozmck> it's made for dispensing. it has a built in controller that has a control language somewhat similar to gcode
[01:54:36] <cradek> for that, getting from one spot to the next fast is important - .002 of slop is no big deal I bet
[01:55:02] <mozmck> could be. I think it's supposed to be accurate to .001
[01:55:29] <cradek> cables directly wound around the stepper shaft is an interesting setup
[01:55:38] <cradek> extremely fast
[01:55:52] <cradek> I've seen that with timing belts too (newer zenbots)
[01:56:32] <mozmck> yeah, this machine moves really fast. I haven't taken mine apart to see exactly how it was setup since it all works.
[02:06:45] <skunkworks> mozmck: I know I might have the old dos version somewhere
[02:07:43] <skunkworks> I think I was able to get it to go into the 700+ipm when I had emc hooked into it
[02:08:20] <skunkworks> but at .015"/step... too coarse
[02:08:40] <cradek> seems like you'd need VERY high accel to ever get up to those speeds
[02:08:43] <mozmck> Mine works as is. I got a basic gui program written to control it with the acl language it uses.
[02:09:21] <mozmck> but the other software might be quite useful...
[02:10:26] <mozmck> my documentation says it does .001"/step - what model machine is yours?
[02:10:46] <skunkworks> I would have to look..
[02:11:13] <mozmck> did you see my pm?
[02:11:27] <skunkworks> look at it this way - the wheel is 1 inch in diameter - so 3.14/200 is around .015"
[02:11:31] <mozmck> mine is a 402G
[02:11:56] <mozmck> maybe it is .001 with microstepping... it may have said something like that in the manual
[02:16:15] <skunkworks> I tried to see how well it positioned microstepping (hoping) but half stepping was about as good as it got.. (around .007)
[02:17:50] <cradek> you can get about 3:1 reduction with one stage of pulleys - but still that doesn't really get you down to what you want
[02:20:22] <elson> Thanks, Chris, for helping me with the sudo! So, I now have the idb component compiling on both systems, and have shown that is does exactly what Anders claimed. It definitely tightens up the response.
[02:21:43] <cradek> cool.
[02:21:52] <elson> You can get 4:1 with XL belts, but it takes a very small motor pulley. 4:1 is no problem with MXL belts with of-the-shelf pulleys. But, the SDB or WM Berg pulleys are a bit pricey.
[02:21:58] <skunkworks> cradek: yes - that was why I kinda figured it wasn't really practical.
[02:22:13] <elson> That's SDP, stock drive products. Geez, my fingers must be cold or something.
[02:24:41] <elson> Yeah, reliable micropositioning in actual systems with friction doesn't get much better than half steps. With a servo, it gets as good as the encoder you can afford. The Cui encoders go down to 2048 cycles/rev, I think.
[02:34:18] <skunkworks> cradek: i am pretty sure they are some sort of ball bearing.
[02:34:23] <skunkworks> bbl
[19:58:12] <alex_joni> micges: sorry, wasn't around
[21:43:41] <CIA-7> EMC: 03micges 07tlo_all_axes * r3abb398cd5fe 10/src/ (4 files in 2 dirs): adapt tooledit.tcl for unified tool format
[21:47:16] <cradek> 11:08:04 -!- Damon [damon@magic.dsndata.com] has joined #designdata
[21:47:30] <jepler> micges: excellent, that's an important step towards making that branch mergeable
[21:47:32] <cradek> oops
[21:48:24] <micges> jepler: thanks to Dewey Garrett
[21:48:56] <jepler> thanks dgarr whereever you are
[21:49:05] <micges> heh
[21:50:07] <micges> jepler: today I realized that fixes to glcanon was VERY important
[21:51:13] <jepler> oh?
[21:51:20] <micges> without it emc would try to run ot of limit without a warning
[21:51:58] <jepler> there are several levels of protection against exceeding soft limits. the axis preview is only one of them, and it's not always accurate
[21:53:50] <micges> but it's doing it's job quite well
[21:56:38] <cradek> I'm convinced that anyone who can write a program in tcl is much smarter than me
[21:58:21] <jepler> if you'd said "write a correct program in tcl", I might agree with you.
[21:58:31] <micges> I have the same feel
[21:59:01] <cradek> yuck, something is wrong with tool touch off in (at least) lathe mode
[21:59:18] <cradek> I think the new offset is correctly written, but the dro doesn't change
[21:59:42] <micges> cradek: sim lathe?
[21:59:46] <cradek> the dro gets updated if you do another g43 or another tool touch off (yes in this case, updated to the value that's "one old")
[21:59:54] <cradek> yes, in master, not tlo_all
[22:00:13] <jepler> ugh
[22:07:46] <cradek> annnnnd now it works
[22:08:03] <skunkworks_> randomly? or did you fix something ;)
[22:08:26] <cradek> randomly
[22:09:07] <skunkworks_> yeck
[22:10:11] <skunkworks_> SWPadnos: those cases you have - do they support a normal hight pci card?
[22:11:10] <cradek> ggrrr
[22:11:28] <alex_joni> * alex_joni has a hunch it doesn't work again
[22:11:34] <cradek> no, it works fine
[22:12:17] <alex_joni> ah, crap.. not beeing able to reproduce it is worse
[22:12:46] <cradek> last night I was using the real lathe and it was not working right - that's why I tried it in sim today.