#emc-devel | Logs for 2007-07-22

[00:52:31] <jmkasunich> Skullworks-PGAB: what is the page?
[00:54:08] <Skullworks-PGAB> Servomate
[00:54:23] <Skullworks-PGAB> does not show on the list
[01:15:47] <jmkasunich> then it doesn't exist yet
[01:16:16] <jmkasunich> why do you think you created it?
[01:28:51] <Skullworks-PGAB> because it started a fesh page with that title up in the yellow top area (where Chips is).
[01:29:16] <Skullworks-PGAB> but at the bottom it still said page was read only
[01:29:50] <jmkasunich> did you enter the password at the preferences page?
[01:30:26] <jmkasunich> that read-only indication isn't on a page by page basis - you can either edit every page, or no page
[01:31:00] <Skullworks-PGAB> yes
[01:31:16] <jmkasunich> very strange
[01:31:24] <Skullworks-PGAB> and when I leave it shows settings saved
[01:31:48] <jmkasunich> go to the main page http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?EmcKnowledgeBase
[01:31:56] <Skullworks-PGAB> I Think I may be having a browser issue
[01:32:01] <jmkasunich> and see if it says "edit text of this page" at the bottom
[01:32:19] <jmkasunich> you gotta have cookies enabled, and you gotta refresh after entering the password
[01:32:29] <Skullworks-PGAB> because the text in the yellow areas does not allways display
[01:32:46] <jmkasunich> you enter the administrator password, not the plain password
[01:33:09] <Skullworks-PGAB> hmm - did not use the admin pass
[01:33:23] <Skullworks-PGAB> that may be it
[01:34:47] <Skullworks-PGAB> ok - thats it - right pass - wrong place
[01:36:07] <Skullworks-PGAB> I work with zombies all week - I fear it is rubbing off on me
[01:37:28] <cradek> Skullworks-PGAB: http://www.asofterworld.com/shirts_asw.htm
[01:37:35] <cradek> Skullworks-PGAB: (see the second t-shirt)
[01:43:29] <a-l-p-h-a_> anyone got cool videos on lathe or cnc work?
[01:43:33] <a-l-p-h-a_> techniques primarily.
[02:10:37] <Skullworks-PGAB> thats a good one
[02:14:14] <Skullworks-PGAB> wooot! I got page 256
[02:55:49] <[1]a-l-p-h-a> [1]a-l-p-h-a is now known as a-l-p-h-a_
[09:42:55] <anonimasu> hm..
[12:54:34] <jepler> cradek: http://emergent.unpy.net/index.cgi-files/sandbox/velaccbug.ngc
[12:54:52] <jepler> cradek: one of petev's bugs (he attributed the problem to failing to home "A") is actually due to this ^^^
[12:54:57] <jepler> maybe you can understand what's wrong
[12:59:59] <jepler> (TRUNK but probably in 2.1 as well -- it's not due to nineaxis, at least I don't think it is)
[13:14:20] <cradek> hi
[13:14:55] <cradek> would you check the one I fixed last night please? I'm not sure I understood it completely.
[13:18:01] <cradek> jepler: which number is wrong in the comment? it says "and max velocity" but I think the calculated velocity is "vel 4.2418e-08"
[13:21:12] <cradek> hmm txyz=0 looks fishy...
[13:21:44] <cradek> but it says cartesian=1
[13:33:24] <jepler> cradek: I'll look
[13:34:13] <jepler> I think txyz was so small that it was less than 0.000001
[13:34:21] <cradek> oh right
[13:34:39] <cradek> if(dx < tiny) dx = 0.0;
[13:34:54] <cradek> I suppose you may have triggered this
[13:35:01] <cradek> const double tiny = 1e-10;
[13:35:10] <jepler> no, my dx is bigger than tiny but still very small
[13:35:10] <jepler> bbl
[13:35:37] <cradek> are you sure after the mm conversion?
[13:36:27] <cradek> yours is e-9 in inch (I guess?) which would be < e-10 in mm
[13:36:59] <cradek> no I'm a dumbass, forget it
[13:43:49] <cradek> jepler: in that case I still don't see which number is wrong
[13:47:11] <jepler> I think I was running the config attached to https://sourceforge.net/tracker/index.php?func=detail&aid=1736182&group_id=6744&atid=106744
[13:47:27] <jepler> the X+A move takes way longer than it should to complete
[13:47:45] <cradek> oh ok
[13:47:58] <jepler> my bug report must not have been very good either
[13:48:00] <cradek> I was looking for a constraint violation
[13:48:02] <jepler> it takes a minute or more
[13:48:06] <cradek> ok
[13:48:08] <jepler> to turn one degree and go a tiny distance
[13:49:23] <cradek> oh I misread "and a (tiny acceleration) and (max velocity)" when you meant "and a tiny (acceleration and max velocity)"
[13:56:18] <cradek> http://pastebin.ca/628287
[13:56:29] <cradek> jepler: ^^ this triggered the bug in linkable()
[13:58:16] <jepler> well that's unusual
[14:01:08] <cradek> what is
[14:01:13] <jepler> the behavior of that gcode
[14:03:12] <cradek> I don't follow
[14:03:16] <cradek> it's drilling holes
[14:05:03] <jepler> the behavior of emc given that gcode -- i.e., the bug
[14:06:05] <cradek> oh, right
[14:06:19] <jepler> my communication skills seem to have atrophied
[14:06:25] <jepler> it must be the french canadian
[14:07:17] <cradek> well I wasn't going to say anything.
[14:08:02] <jepler> * jepler ponders a quadrature -> step+direction converter on AVR
[14:08:49] <cradek> I think we could do those short step pulses (one per period) in stepgen and get the same benefit
[14:09:07] <jepler> that needs collaboration between stepgen and the hardware driver
[14:09:35] <jepler> there's also Thomas Fritz's circuit which does one step per edge of the step signal
[14:09:45] <jepler> that just needs a stepgen modification, a new step type
[14:09:47] <cradek> his looks interesting
[14:10:05] <jepler> I wonder if he's made that modification, or if this is just an idea right now
[14:10:12] <jepler> (that post is why I'm thinking about it again)
[14:10:15] <cradek> and you could still see the output in halscope with his, not with mine
[14:11:39] <jepler> the alternative that doesn't need hardware seems nicer
[15:17:07] <jmkasunich> morning guys
[15:17:09] <jepler> hi jmkasunich
[15:17:30] <jmkasunich> ISTR that Mariss of gecko once posted a very simple circuit that could turn quadrature into step/dir
[15:17:52] <jmkasunich> he used RC delays, a quad XOR gate, and a D flip-flop, IIRC
[15:18:05] <jmkasunich> I'll try to find or recreate it
[15:18:55] <jepler> one of my electronics books has a quadrature to step+dir converter, but it does nothing to guarantee direction setup/hold times
[15:19:13] <jmkasunich> I suspect that would be true of Mariss's circuit as well
[15:27:10] <alex_joni> hi guys
[15:29:30] <jepler> hi alex
[15:31:31] <SWPadnos> hi guys
[15:35:05] <alex_joni> jepler: still in canada?
[15:35:13] <alex_joni> hey SWPadnos
[15:36:59] <jepler> alex_joni: nope, I'm home
[15:37:03] <jepler> hi SWPadnos !
[15:37:18] <SWPadnos> hey alex_joni and jepler
[16:41:15] <SWPadnos> has anyone heard any RT reports for specific (recent) chipsets?
[16:42:01] <SWPadnos> my wife's Shuttle XPC seems pretty good, it's based on the Nvidia NForce 570 Ultra
[16:47:25] <jepler> SWPadnos: the machine where I was doing rtai 64bit is "nVidia Corporation MCP51" according to lspci
[17:13:32] <SWPadnos> thanks. I'll have to check that again when I do more testing (later today, I think)
[17:50:39] <jepler> the motherboard is: GIGABYTE GA-M51GM-S2G Socket AM2 NVIDIA GeForce 6100 Micro ATX AMD Motherboard - Retail
[17:55:01] <SWPadnos> hmmm. that's pretty similar to the Foxconn board I tried, which was absolutely horrendous
[17:57:43] <jepler> I haven't tried the official 32bit kernel
[17:58:51] <cradek> jepler: I fixed that problem
[17:59:09] <cradek> may still be a problem at 1e-15 that I left there for superstitious reasons.
[18:00:45] <SWPadnos> strange - I'd think that 1e-10 would be small enough since floats would tend to lose precision there
[18:00:57] <SWPadnos> (and HAL uses floats, even if canon uses doubles)
[18:01:26] <cradek> I wish I had the gcode that caused me to put in the 'tiny' check, but I don't
[18:06:19] <jepler> hmm
[18:06:30] <jepler> without the patch, I have a gcode that causes a following error on that configuration
[18:06:35] <jepler> G0 X0 A0 Y0Z0
[18:06:36] <jepler> G0 X.00001 A1
[18:06:39] <jepler> haven't tried with the patch yet
[18:07:03] <jepler> I'm not sure what causes a following error in servo-sim, maybe the config is not right or something
[18:07:05] <jepler> (it's from petev)
[18:07:08] <jepler> bbiab
[22:58:20] <SWPadnos> hmmm. so here's an interesting piece of latency data
[22:58:39] <SWPadnos> with the PC apparently crashed, the latency is much much better
[22:59:07] <SWPadnos> the squarewave I'm looking at with a (real) scope has edges consistent to within 1 uS
[22:59:28] <cradek> ha
[22:59:58] <SWPadnos> I had left the machine doing updates, when I just looked at it, the chemical model screensaver was onscreen, but the chargepump/mesa drivers are still doing their thing
[23:00:10] <SWPadnos> and of course it doesn't wake up when I type or move the mouse
[23:01:01] <cradek> do you know what's wrong with it?
[23:01:18] <SWPadnos> not yet
[23:01:35] <SWPadnos> I didn't manually enable ssh, so I'm not sure I'll be able to tell remotely
[23:01:46] <SWPadnos> (ISTR that ssh is off by default)
[23:01:50] <cradek> yeah
[23:03:38] <SWPadnos> it doesn't respond to pings
[23:04:00] <SWPadnos> interesting that the kernel can be that hosed, but RTAI still runs
[23:11:47] <Skullworks-PGAB> thats a good thing ( I think )