#emc-devel | Logs for 2010-09-16

[00:35:47] <skunkworks> KimK: ! how is it going?
[00:36:13] <skunkworks> maybe emc could do something like this? ;) http://groups.yahoo.com/group/mach1mach2cnc/message/121997
[00:37:04] <cradek> well, he spelled hobbyist right
[00:41:59] <skunkworks> heh - (I guess I cannot make jokes about that)
[00:43:06] <skunkworks> dad has most of the adaptors machined up for the y axis. Now we need to see how things line up ;)
[00:43:24] <cradek> yay
[00:43:56] <skunkworks> if we did everything right - it should just couple up.
[01:34:06] <Dave911> skunkworks: >>maybe emc could do something like this?
[01:34:08] <Dave911> ...... hmm.... 2 hrs have gone by no one has muttered a word about that idea on the Mach3 list ... support has been downright underwhelming.. :-) I suspect that this idea and the concept of voluntary income tax are probably dead ideas....either that or the diehard Mach3 fans are attempting to drown their sorrows at the local tavern..
[01:57:54] <skunkworks> Dave911: yes. I think you are right
[02:02:45] <skunkworks> Dave911: http://www.machsupport.com/forum/index.php/topic,13294.msg107181.html#msg107181
[02:17:29] <Dave911> Did you see the last reply from Ray L. He says he heard from Brian and V4 would be out soon - that message was posted on 9/13/10. Yesterday, 9/14/10 Brian dropped the bomb that Mach3 V4 was on indefinite hold. The earliest promised release date for V4 that I can recall was April of 2009. However it was actually probably promised before that ... I would have to go back through my...
[02:17:31] <Dave911> ...notes. Brian said that he was going to stop promising V4 delivery dates about March of this year. I gave up on Mach3 in the July-August 2009 time frame and I recall thinking at the time that I wasted at least a year trying to deal with Mach3 issues that were simply hardcoded errors. I personally explained several of the errors to both Brian and Art. I even wrote up a 2 plus page...
[02:17:32] <Dave911> ...report on one serious timer issue I found. The windows timers were really screwed up by Mach3 and they were no longer functioning as designed. When Brian said he would look into it and then never did, even when I supplied sample code, I threw in the towel. I was writing a plug in for Mach3 and with the timer issue I simply could not make the plug in work. So in effect I was stuck...
[02:17:34] <Dave911> ...until the Mach3 issue was fixed. It never was.
[02:21:40] <morficmobile> this sounds a lot like MachNukem, sorry, just sayin
[02:40:58] <Dave911> NP.. unfortunately a lot of people have spent a lot of time trying to get Mach3 to work..and for a long time Art and to some extent Brian were making headway.. then for whatever reason it all came grinding to a halt...
[02:51:26] <skunkworks> Dave911: you have a good attitude. Others that make the transistion just complain about what emc doesn't do compared to mach.
[02:51:47] <skunkworks> not noticing that it is a great, stable control.
[02:52:15] <skunkworks> (and probably why they changed)
[03:01:02] <Dave911> EMC2 is amazing software..extremely stable. >>>Others that make the transistion just complain about what emc doesn't do compared to mach.
[03:01:03] <Dave911> That is only because they have no idea how broken Mach3 is. There really is no comparison..
[03:07:01] <morficmobile> Dave911: if he is a one man show, who knows what happened IRL that caused him to fall behind
[03:11:23] <Dave911> True ..however I used to talk to him frequently for quite a while and I could almost see the wheels coming off a few times. Pretty disheartening when you can sense things going in the wrong direction but really can't do anything about it.
[03:12:45] <Dave911> Life goes on.. fortunately while I was headed in the wrong direction with Mach3, the EMC2 devs were hard at work moving EMC2 along :-)
[03:20:55] <CIA-5> EMC: 03cradek 07v2.4_branch * ra245338c5976 10/src/emc/rs274ngc/interp_convert.cc: Fix incorrectly scaling ABC tool lengths by 25.4
[03:23:48] <cradek> I doubt anyone will ever use that, but it might as well work right
[03:31:18] <morficmobile> does this mean it did not scale (correctly) by 25.4 and now it does?, or that it was doing it when it didn't need to?
[03:33:35] <cradek> those are tool "lengths" that apply "along" the rotary axes. they're in degrees so they should not scale according to inches/mm
[03:33:58] <cradek> maybe useful for a puma robot or something - I dunno
[03:46:49] <morficmobile> degrees is the only thing i did not have to "relearn" coming to the US, yeah :)
[03:49:14] <cradek> haha
[08:09:42] <alex_joni> yeah degrees are fun, I remember UNITS = 1 ;)
[08:37:58] <micges_work> proposed patch for master improving homing error messages: http://codepad.org/8piHJGaK
[09:05:54] <alex_joni> looks good for me
[09:11:32] <morfic> i wonder how much face palming it'll cause when people wake up and try to reread the #emc dialog there
[09:38:30] <alex_joni> do you really think somebody reads that ? :P
[09:51:40] <morfic> some people are anal about reading their backlog, others even end up in the middle of it all, wondering why they did it ;)
[18:26:16] <alex_joni> http://www.youtube.com/user/ali0223#p/a/u/0/aPv_FE2qwA0
[18:29:02] <jepler> they need an anti-vibration tripod
[18:35:53] <Jymmm> Ah, the red and green text (kanji) is saying when the anti-vibration is on/off
[18:36:01] <jepler> that was my guess
[18:36:33] <Jymmm> Now, WHAT that is I wish I knew =)
[18:43:16] <tom3p> when fuzzy logic 'founder' went to japan for celibration of his work, the group showed off a motor with a plate balanced on top of a rod. he left, returned, placed a live mouse on the plate, and the fuzzy logic kept plate in balance even with freaked out mouse.
[18:46:04] <alex_joni> holy crap: http://www.youtube.com/watch?v=v5eR0eHknZk
[18:46:43] <tom3p> not kanji, chinese (traditional) from taiwan http://www.hofo.com.tw/
[18:49:54] <alex_joni> this is potentially faster: http://www.youtube.com/watch?v=0-Kpv-ZOcKY&feature=related
[18:52:53] <andypugh> This is faster, I think.
[18:52:55] <andypugh> http://www.youtube.com/watch?v=Gv5B63HeF1E
[18:53:14] <alex_joni> this is cooler: http://www.youtube.com/watch?v=cJuNe50uuzk&feature=related
[18:57:43] <tom3p> the kingservo manual suggests the dampening is a stock feature
[19:10:33] <tom3p> ah, the stock name for the plate on a stick balance is 'the inverted pendulum' http://www.iiisci.org/journal/CV$/sci/pdfs/S406IQ.pdf
[19:18:42] <tom3p> wow, hate those pay for scientific papers? try http://www.doaj.org directory of open access journals
[19:23:04] <jepler> nice. I was looking for a paper earlier today (on a certain RNG algorithm) and found a paid version http://www.jstor.org/pss/2959748 but then quickly turned up a free version http://projecteuclid.org/DPubS?service=UI&version=1.0&verb=Display&handle=euclid.aoap/1177005878
[19:23:56] <jepler> as I'm not an academic I'm not likely to pay for a paper..
[19:27:35] <Jymmm> jepler: Are you trying to do it in software?
[19:28:45] <Jymmm> jepler: As in, do you want a PRNG or a TRNG ?
[19:28:52] <jepler> I read that the next C++ standard may include a random number generator based on that paper
[19:28:58] <jepler> as I'd never heard of the method, I went to find out about it
[19:29:26] <Jymmm> jepler: Oh, ok. then nm. Thought you were looking for a TRNG
[19:30:16] <Jymmm> jepler: Yeah, that method has been around for a few years.
[19:30:48] <Jymmm> jepler: They are jsut using the timing of RAM =)
[19:41:52] <Jymmm> jepler: http://www.iro.umontreal.ca/~simardr/testu01/tu01.html
[19:50:14] <Jymmm> jepler: This isn't the exact link I was looking for, but it should help explain... http://www.sciencedaily.com/releases/2010/02/100222082529.htm
[19:51:05] <Jymmm> The German team has now developed a true random number generator that uses an extra layer of randomness by making a computer memory element, a flip-flop, twitch randomly between its two states 1 or 0. Immediately prior to the switch, the flip-flop is in a "metastable state" where its behaviour cannot be predicted. At the end of the metastable state, the contents of the memory are purely random.
[19:51:17] <Jymmm> That's the jest of it
[19:53:17] <Jymmm> jepler: Personally, I still like this... http://www.fourmilab.ch/hotbits/hardware3.html
[20:06:08] <jepler> earlier this year I built a TRNG based on the reverse-biased junction, similar to this: http://www.entropykey.org/tech/
[20:06:21] <jepler> much less exotic
[20:08:05] <jepler> (though my device only has a single junction and doesn't do as much analysis of the produced random stream)
[20:08:11] <Jymmm> jepler: Yeah, I just like that I can say I have radioactive material =)
[20:11:19] <Jymmm> jepler: I mean come on... how many times do you get a chance to use "CAUTION RADIOACTIVE EXPOSURE" stickers for real =)
[21:51:48] <andypugh> Is it possible to have a 4-joint 3-axis machine?
[21:52:52] <andypugh> spiderdab and a chap on the forum with a gantry are both struggling with a similar issue, which seems to be related to a spurious A-axis, without which you can't have 4 joints in joint mode.
[21:54:51] <andypugh> the stepper-gantry example config has the same issue.
[21:55:48] <cradek> andypugh: I think I'd just do it in hal and hide the extra joint from emc
[21:56:36] <andypugh> what is "emc" in this context?
[21:56:50] <cradek> motion/kins
[21:57:34] <alex_joni> andypugh: you can in ja3, but not on 2.4.x
[21:58:20] <andypugh> Set up as a 3-axis and manually add a stepgen you mean? But then there are only 3 joints to jog in Axis?
[22:00:17] <andypugh> I am guessing that Spiderdab with his tetrapod needs to jog each wire individually to home, so he needs 4 joints in joint mode. In his case he might be best off just ignoring the A axis in World mode?
[22:02:11] <andypugh> (Slow-moving brain recalls this discussion from last time). You are saying to manually set num_joints to 4 and not to take it from [TRAJ] AXES?
[22:02:48] <cradek> maybe he should try running ja3 branch
[22:08:40] <andypugh> The message is on the mailing list from Spiderdab, perhaps someone who knows more about the ja3 branch can advise?
[22:19:23] <cradek> am I losing my mind thinking jmk made motion use the new free_tp?
[22:24:31] <andypugh> There are free_tp pins
[22:24:40] <andypugh> But I don't know what they do
[22:27:14] <cradek> ah, I was thinking of 15df52f308b, which is not merged
[22:28:13] <andypugh> It seems that in 2.4.3 a 3-axis ini with a 4-joint motion doesn't work either, the 4th joint doesn't seem to get any motion.pos-cmd unless it exists in the INI too.
[22:33:38] <alex_joni> cradek: I wouldn't suggest anyone to run ja3 right now
[22:36:15] <andypugh> Gah! I was so long expeimenting that I got logged out of the forum, so my long reply was lost in the aether when I tried to post it.
[22:37:05] <alex_joni> forums suck
[22:38:14] <JT-Hardinge> the limit is 2hrs atm andypugh
[22:38:53] <andypugh> I might have logged in from another window or something then. The bizarre thing is is knows who I am, and list me by name, not as a guest.
[22:39:17] <JT-Hardinge> I bumped it up to 4 hrs for you long winded guys :)
[22:39:41] <JT-Hardinge> do you log in on the left side of the panel?
[22:43:55] <andypugh> Yes
[22:44:21] <andypugh> But only when I realise I am logged out. And I am pretty sure it wasn't a timeout.
[22:48:20] <JT-Hardinge> forums suck then
[22:48:30] <JT-Hardinge> :P
[22:53:09] <andypugh> fascinating; Even if [AXIS_3] is TYPE=LINEAR its jog speed is still governed by the rotary jog speed in an XYZA config. I think that counts as a bug?
[23:03:08] <jepler> in the sense of something I'd be eager to fix in 2.4, no. "A" is always a rotary axis. TYPE=LINEAR has something to do with joint mode, but improving nontrivkins behavior is not "bugfixing" for the stable branch.
[23:03:15] <jepler> IMO, of course
[23:03:19] <jepler> IMHO, I mean
[23:04:30] <andypugh> A is a rotary axis, yes. But why does that make joint 3 a roatary axis?
[23:05:15] <andypugh> (Yes, I know it is all a bit tangled up)
[23:05:45] <andypugh> I am trying to make an XYZU config, but so far with no luck.
[23:06:07] <andypugh> (even with 7 joints and 7 sets of [AXIS_N] stanzas
[23:07:26] <andypugh> Ah, OK, I have it now. Let me fiddle some more
[23:31:15] <andypugh> XYZU with 7 axes and [AXIS_6] identical to [AXIS_3] seems to work as far as creating a joint where the speed is controlled by the linear jog speed is concerned.