#linuxcnc-devel | Logs for 2015-09-11

[00:31:26] <KimK> The 2.7 ISO download address ( http://www.linuxcnc.org/linuxcnc-2.7-wheezy.iso ) seems to be running very slow on zsync here. Wget is a little better, but not a lot. Bpth stall frequently. Is anyone else seeing this trouble, or is it just me? (This is from the getting 2.7 directions at http://linuxcnc.org/docs/2.7/html/getting-started/getting-linuxcnc.html )
[00:32:33] <KimK> s/Bpth/Both/
[01:03:46] <seb_kuzminsky> zsync is making ~1.5 MBps here
[01:04:50] <KimK> OK, I'll keep looking into it then. Thanks, Seb!
[16:02:43] <skunksleep> http://www.cnczone.com/forums/vertical-mill-lathe-project-log/281738-cnc-cad.html
[16:06:04] <cradek> ... so I hear this guy spent 2000 hours trying to cut threads with mach3
[16:06:34] <malcom2073> Definition of insanity: Keep trying the same thing, thinking the results will change
[16:06:59] <jepler> you mean this specific comment, then? http://www.cnczone.com/forums/vertical-mill-lathe-project-log/281738-cnc-cad.html#post1759368
[16:07:14] <cradek> oh, haha, page 2 is where it gets funny
[16:07:41] <malcom2073> LOL
[16:07:45] <malcom2073> Wow... someone get a needle and pop that head
[16:08:05] <jepler> > Show me some leadscrews, done to a tolerance of 2 microns or less, over 400 mm.
[16:08:13] <jepler> that would be a nice output
[16:08:44] <cradek> that would take very good hardware (running linuxcnc)
[16:09:05] <cradek> but you do that with a grinder, not a threading tool
[16:10:01] <PCW> Yeah and probably quite slowly
[16:10:28] <cradek> he last used emc in 2002, so he knows it can't cut good threads
[16:10:45] <cradek> haha that's really great
[16:11:42] <jepler> PCW: congrats for trying to engage with this guy twice, double congrats if you can avoid engaging with him a third time in this thread.
[16:11:53] <PCW> needs good mechanics and perhaps a linear scale nothing really control related if the control gets the gearing right
[16:11:53] <cradek> my thoughts exactly
[16:12:05] <PCW> Yeah Im done
[16:12:16] <skunksleep> When was threading added?
[16:12:31] <cradek> skunksleep: don't bother, go back to sleep
[16:13:45] <cradek> near March '06
[16:14:17] <skunksleep> Oh - at this point I don't get involved
[16:14:35] <cradek> heh
[16:14:38] <skunksleep> Probably..
[16:14:45] <cradek> so, not 10 years ago, but darn close
[16:14:46] <PCW> I suspect that good spindle speed control is also needed at least to avoid Z tracking errors as much as possible
[16:15:09] <jepler> 2micron/400mm = 5ppm < heat expansion coefficient of steel per °C
[16:15:53] <PCW> Yeah that a rather special machine to do that, not the machine the OP was designing
[16:16:01] <PCW> that's
[16:16:30] <jepler> I don't know about "aerospace 3a/b specs" either but it seems entirely possible that the mechanical characteristics of >99% of lathes might not be up to it either, before you talk about software
[16:19:15] <PCW> https://www.engineersedge.com/thread_strength/thread_classes.htm
[16:19:49] <ssi> that's stupid, thread class has nothing to do with the precision of the machine making it, just has to do with what pitch diameter you shoot for
[16:23:17] <ssi> my junky stepper-driven G0602 can cut class 3 threads no problem
[16:24:15] <jepler> I stand corrected
[16:24:37] <ssi> jepler: not aiming to correct you, just mr. "professional" in thailand :P
[16:25:10] <jepler> I had no idea whether these kinds of thread classes he was talking about were as unreasonable as a the leadscrew he mentioned first
[16:25:47] <ssi> if you're making micrometer barrels, you might can convince me that your average DIY cnc lathe isn't accurate enough to cut the threads
[16:25:50] <ssi> maybe
[16:27:15] <jepler> we could bet him the cost of the gages that you can get a class 3 screw off your lathe or cradek off his sherline
[16:27:35] <ssi> lol
[16:27:46] <cradek> not a 400mm thread because that's longer than the whole lathe
[16:28:09] <skunksleep> Hardidge
[16:28:10] <cradek> I could send him one of my 90tpi screws to measure
[16:28:19] <ssi> if he needs to make a 400mm long class 3 thread, he probably doesn't need to be diy'ing a lathe for it
[16:29:35] <cradek> eh, I don't care about this anymore. thanks for the entertainment, though
[16:29:47] <ssi> lol
[16:31:42] <skunksleep> On a lighter note - I won a kindle fire hd from my cable company.. Random
[16:32:48] <skunksleep> Didn't know of the drawing.. Got entered because we auto-pay.
[17:04:52] <skunkworks> PCW, being lazy - is there a 5i25 mx3660 config?
[17:05:02] <PCW> Yes
[17:05:37] <PCW> I think its a mx3660x2 config
[17:05:50] <skunkworks> ok. what about the 7i92?
[17:08:09] <PCW> 7i92_MX3660x2D.bit
[17:08:31] <skunkworks> awesome!
[17:14:46] <PCW> D means DPLL
[17:16:09] <skunkworks> I should be able to test that on a real system.
[20:45:37] <mozmck> I might have found a bug in 2.7, or I might be ignorant of some settings I need to tweak.
[20:46:49] <mozmck> I have some code with an M67 in between straight lines, and it makes motion pause for a bit. http://pasteboard.co/zzTUBUc.png
[20:49:49] <mozmck> Here is the code behind that velocity plot: http://pastie.org/10413572
[20:52:10] <mozmck> When I move the M67 farther from the end of the line, the dip in velocity is smaller. Notice also that there is no dip except on the last M67.
[20:52:53] <mozmck> Any ideas? Are there some TP settings that might change that behavior?
[20:58:52] <skunkworks> i have tested it recently and it seemed to work
[21:07:43] <PCW> its not _supposed_ to be a queue buster
[21:08:22] <mozmck> Yeah, but it's not quite doing right.
[21:25:23] <skunkworks> mozmck: you know G0 to G1 or vice versa is exact stop?
[21:27:36] <mozmck> I'm not sure what that means?
[21:28:51] <mozmck> Obviously motion does not and should not stop between each line segment.
[21:29:49] <skunkworks> what line are you seeing the pause?
[21:30:23] <mozmck> Line 50 and line 58
[21:30:41] <mozmck> At the M67 just before the last G1 line segment.
[21:32:16] <mozmck> I guess you mean that a G0 move must come to a complete stop before a G1 move starts and vice versa? That is not where I'm seeing the glitch.
[21:34:42] <skunkworks> ok
[21:36:33] <mozmck> I can run that code here in simulator mode and see the same glitch.
[21:38:55] <skunkworks> I see it. the M67 seems to cause a dip in velocity
[21:39:01] <mozmck> yes.
[21:39:05] <skunkworks> *The last
[21:39:30] <mozmck> Yep! We noticed it while plasma cutting today.
[21:39:49] <skunkworks> I think that requres a bug report
[21:40:30] <mozmck> Ugh.
[21:41:28] <mozmck> I played with making the last line segment longer and shorter, and when it is longer the dip is much smaller.