#emc-devel | Logs for 2007-09-19

Back
[00:34:58] <skyfox00> * skyfox00 has returned in full force.
[04:52:56] <SWPadnos> ah-ha! I found the userspace <-> kernel communications interface - it's called connector
[04:53:30] <SWPadnos> it's in kernel config, under drivers / connector. called the "unified userspace <-> kernelspace linker"
[04:54:18] <SWPadnos> it works on top of netlink. it may be useful for implementing kernel command handlers for userspace code
[09:11:06] <alex_joni> http://www.mjmwired.net/kernel/Documentation/connector/
[12:08:59] <cradek_> cradek_ is now known as cradek
[12:12:35] <jepler> good morning cradek
[12:12:42] <cradek> hi
[12:13:28] <jepler> I'm sure puzzled by skyfox00's symptoms
[12:13:50] <cradek> I haven't read back yet
[12:14:16] <jepler> it's as though all the inputs work but all the outputs are broken
[12:14:31] <jepler> 18:41:32 <skyfox00> ok, dout-00 has a LOW and dout-01 has a HIGH regardless of setp pluto-servo.dout-0[1/2] true/false
[12:14:52] <jepler> and broken to set values either, as the watchdog would do
[12:15:32] <cradek> maybe he got a bad pluto?
[12:18:15] <jepler> back on 08-30 he did get the LED to blink with the pluto firmware -- http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-08-30.txt
[12:18:46] <jepler> beats me
[12:20:00] <cradek> could his power supply be wrong or bad?
[12:20:06] <jepler> once again, beats me
[12:20:27] <jepler> 18:01:32 <skyfox00> and touching the Q[ABZ]* pinheaders causes the counters to jump around then return to zero...
[12:20:37] <jepler> seems like this is a pretty risky way to test whether inputs are working :-P
[12:21:02] <cradek> bbl
[13:17:37] <alex_joni> bbl
[13:26:51] <jepler> bbl
[13:26:52] <jepler> ?
[14:17:58] <jepler> good morning skunkworks_
[14:23:21] <skunkworks_> Good morning. Wine tasting on the island girl (boat) last night. Bit sluggish this morning.
[14:26:16] <jepler> is that what the _ means
[14:26:20] <jepler> ?
[14:26:33] <skunkworks_> mental note - eat something before going on a wine cruise.
[14:27:06] <skunkworks_> heh - no. I have a computer on at home on logged into irc.
[14:30:48] <SWPadnos> uh-oh. another _
[16:14:15] <cradek> does anyone think they understand this discussion about the touch off screen?
[16:16:51] <fenn> sounds like the users are entering non-numeric characters and something goes wrong
[16:17:23] <cradek> it grays out the OK button (and tells you why) whenever the entry is invalid
[16:17:39] <cradek> (that's a feature)
[16:18:09] <SWPadnos> I usually use <home>, <shift>-<end>, then type in the number (in other programs that have the same problem)
[16:18:41] <SWPadnos> does the dialog select all the edit text when it pops up?
[16:18:47] <cradek> so you think they get something crazy in there and don't know how to delete it?
[16:18:52] <SWPadnos> yes
[16:19:00] <SWPadnos> even a leading space can be a problem, I think
[16:19:14] <cradek> I've wondered if people just see it grey out and freak, not understanding what's going on
[16:19:17] <jepler> but each time the screen is shown it's supposed to be filled with "0.0" and all text selected
[16:19:18] <SWPadnos> could be
[16:19:36] <jepler> so when they dismiss it and bring it back, OK should be ready to click
[16:19:41] <SWPadnos> if the text is all selected when the dialog appears, then it's probably something different
[16:19:46] <cradek> I wish I knew if there was a bug, or if the design could be improved
[16:19:53] <SWPadnos> heh
[16:20:01] <SWPadnos> gotta run for a bit - just thought I'd stick that out there
[16:20:03] <cradek> I don't know which it is
[16:20:42] <cradek> we had another report a while back about it doing something strange, but I never understood what it was, and I don't remember if/how it was resolved
[16:20:44] <jepler> the number is fed to the gcode interpreter, so stuff like extra spaces should be accepted (and is in my testing)
[18:17:17] <SWPadnos> crap. that's something I think we'll need to think about soon. people using EMC on non-net-connected machines, and how they get updates
[18:18:07] <SWPadnos> jepler, regarding the touch-off dialog, there must be something else happening before the text is passed to the interp, or the dialog wouldn't be able to tell you the number is invalid
[18:22:04] <skyfox00> jepler: after you left last night I reported the results of my testing of pluto... you see it?
[18:22:13] <jepler> skyfox00: no, let me scroll back
[18:22:28] <jepler> 18:41:32 <skyfox00> ok, dout-00 has a LOW and dout-01 has a HIGH regardless of setp pluto-servo.dout-0[1/2] true/false
[18:22:32] <jepler> 18:42:16 <skyfox00> they do not seem to be Z though..
[18:22:35] <jepler> oh, this?
[18:22:36] <skyfox00> yeah, that.
[18:22:43] <jepler> I can't understand the behavior you're seeing
[18:22:58] <skyfox00> I cant either.
[18:23:40] <skyfox00> well, I'll try the latest cvs-version and see how that works, etc.
[18:23:56] <jepler> you're sure you were testing the right pins?
[18:25:11] <skyfox00> I think so...
[18:25:16] <jepler> am I right in thinking that you did get the LED to ramp on a different system with the Live CD? If so, can you recreate that situation? It could rule out damage to the board.
[18:26:27] <skyfox00> yeah, in the past I did get the LED to fade in and out on the fast computer with the nvidia card running a buggy kernel and the cvs version of emc!
[18:27:17] <SWPadnos> can you boot from the LiveCD on the computer you're using now?
[18:29:46] <skyfox00> I can boot the liveCD on the 350Mhz ok, but it takes forever to load(only 128MB ram).... when I tried to run the pluto.hal file to see if the LED would blink, it bombed out because the halfile was modified to work with the cvs version(dots v. dashes) and then the parport was hanged and I didnt want to reboot again..
[18:30:50] <skyfox00> thats the only reason I want the cvs version, it can recover from EPP errors...
[18:30:55] <SWPadnos> is there a pluto demo config on the liveCD? (I don't track pluto development closely)
[18:31:27] <skyfox00> there is a lathe pluto config, but no LED blink demo
[18:31:48] <jepler> ... running lathe-pluto and jogging the X axis should also make the LED turn on and off
[18:31:56] <jepler> well, except it'll ferror real quick
[18:32:08] <skyfox00> then the led will go off?
[18:32:52] <jepler> yeah emc will stop trying to drive the X axis servo as soon as it sees it's not getting encoder feedback
[18:34:22] <skyfox00> well, I could just do halrun -I and then loadrt pluto_servo then start doing setp's, right?
[18:34:27] <jepler> yeah sure
[18:34:32] <jepler> and addfs and so forth
[18:34:48] <skyfox00> ok, what did addfs do agian?
[18:35:19] <skyfox00> ok, I just read, for adding threads
[18:35:20] <jepler> unless you arrange for the "write" function to be executed periodicaly, nothing will happen
[18:35:29] <skyfox00> got it.
[18:35:39] <jepler> so you create a thread with 'loadrt threads', add functions to the thread with 'addf', and tell the threads to start executing with 'start'
[18:35:50] <jepler> all these things are (should be) in the .hal file you're using now
[18:36:30] <skyfox00> yeah, buts its the cvs-version syntax...
[18:36:39] <skyfox00> but I think I can manage..
[18:36:55] <SWPadnos> skyfox00, meet nano. nano, meet skyfox00
[18:37:03] <skyfox00> ?
[18:37:11] <SWPadnos> nano is a simple text editor :)
[18:37:24] <SWPadnos> or you can use gedit or whatever
[18:37:32] <skyfox00> oh, yeah, I use mc/mcedit and pico for other things...
[18:38:06] <skyfox00> I know how to use a text editor, I wrote a moving map display program in one.
[18:38:24] <skyfox00> in C, that is.
[18:38:34] <SWPadnos> heh :)
[18:38:58] <skyfox00> hey, its the best linux map program I have!
[18:39:28] <SWPadnos> map like GPS / terrain, or map like dynamic network map? (or something else)
[18:40:05] <skyfox00> ok, yeah, reads data from a GPS and then displays B&W 1M/Pixel photo's
[18:40:35] <skyfox00> does routes/waypoints, etc....
[18:40:39] <SWPadnos> nice
[18:41:32] <skyfox00> I actualy open sourced it for trade for a 1.33 Ghz toshiba portege 3500
[18:41:38] <skyfox00> * skyfox00 grins
[18:42:16] <skyfox00> ok, I'm gona go try some stuff... thanks for all the help
[18:42:28] <SWPadnos> enjoy (hope the plutoo isn't shot)
[18:42:31] <SWPadnos> -o
[18:42:37] <skyfox00> me to!
[18:42:47] <skyfox00> * skyfox00 is away: for now...
[19:13:36] <skyfox00> * skyfox00 is back
[19:15:20] <skyfox00> ok, I just tested pluto on the 350Mhz using the liveCD and the behavior is the same, the LED never blinks and touching the encoder inputs makes the counters jump and ecoder 0 always reverts back to 13 after a few seconds...
[19:15:42] <alex_joni> doesn't sound too good
[19:15:51] <cradek> what power supply are you using?
[19:16:24] <skyfox00> 5V 1000mA
[19:16:40] <skyfox00> its the same one I was using befor that worked fine...
[19:17:23] <fenn> does the pluto board come with any test software?
[19:17:52] <SWPadnos> check it the same way on the PC that worked before
[19:17:52] <skyfox00> come to think of it, yeah, I will try that...
[19:21:07] <jepler> fenn: yeah, it comes with a few sample firmwares & windows loader app
[19:22:20] <fenn> * fenn thinks skyfox00 might need to learn how to de-solder chips
[19:23:22] <skyfox00> LEDblink.rbf makes the LED blink!
[19:23:54] <cradek> skyfox00: are you plugging the pluto directly into the parport, or through a cable of some kind?
[19:24:22] <skyfox00> I learned my lesson about using cables a long time ago ;)
[19:24:33] <skyfox00> um, straight into the computer
[19:25:04] <jepler> LEDblink.rbf doesn't use EPP communication though
[19:25:42] <skyfox00> no, and I'm running it useing the windows FPGAconf.exe on a pnasonic CF-25 166Mhz that does not do EPP
[19:26:15] <skyfox00> my point is the pluto can still blink the LED
[19:26:16] <cradek> which version of emc are you testing with?
[19:27:08] <jepler> the live CD would be emc 2.1.6 I think
[19:29:36] <skyfox00> next I plan to load the liveCD up on a computer that has worked befor...
[19:32:35] <skyfox00> oh rats, I forgot the liveCD in the other building... brb
[19:33:22] <skunkworks_> * skunkworks_ has about 15 coppies laying around
[19:34:18] <skyfox00> oh, as to unsoldering chips, just strap the board down and use a heat gun untill the solder melts and then pull the chip in question of with tweazers...
[19:34:39] <skyfox00> off with tweazers.
[19:35:11] <skunkworks_> what virtualization does the compile server use?
[19:35:19] <jepler> skunkworks_: vmware I think
[19:35:26] <jepler> the gratis vmware server, to be more specific
[19:35:41] <skunkworks_> I thought vmware was a pay software? no?
[19:35:48] <skunkworks_> * skunkworks_ googles
[19:36:02] <cradek> you can get a free-of-charge version now
[19:36:35] <cradek> it's a minor pain to set up and run
[19:36:43] <skyfox00> My brother designes circut boards for a living and has to solder down 40 pins/inch somtimes(with a sterio-microscope) and so If I need to replace a chip....
[19:36:45] <skunkworks_> ok
[19:37:57] <jepler> there's another free vmware product called "vmware player"; I don't understand what the differences are
[19:38:32] <skyfox00> also, If you get a board hot and tap it on the bench, ALL the chips will come off.
[19:39:15] <skyfox00> works well for the mass slavedge of parts from old junker boards...
[19:41:00] <skyfox00> well, the music played...
[19:42:05] <skyfox00> ok, it booted, time to crawl down behind the computer and plug pluto in ;0
[19:42:11] <alex_joni> did the fat lady sing?
[19:42:26] <skyfox00> Ubuntu startup music.
[19:42:35] <alex_joni> then it's not over yet
[19:42:45] <alex_joni> it's not over till the fat lady sings
[19:43:25] <skyfox00> Yuck! its dusty down there...
[19:44:40] <skyfox00> does Ubuntu have a feature I dont know about?(I did have to unplug the speakers so I could plug in the usb thumbdrive...)
[19:45:29] <alex_joni> skyfox00: sorry, that was a quote from some movie
[19:45:57] <skyfox00> ok, pluto works fine, the LED fades in/out and all the counters started up a 0.
[19:46:08] <skunkworks_> sweet
[19:46:19] <alex_joni> probably a bad programming last time
[19:46:32] <skyfox00> so the 350Mhz just cant cut the mustard
[19:47:18] <skyfox00> alex_joni: even with the liveCD, pluto did not work right on the 350.
[19:47:55] <skyfox00> jepler: you still here?
[19:48:15] <SWPadnos> hmmm. Subversion uses binary deltas
[19:53:50] <skyfox00> well, looks like I need to find a new host for pluto.
[19:54:15] <SWPadnos> so the liveCD works on the other PC but not on the slow one you want to use?
[19:57:55] <skyfox00> nothing works on the 350, everything works on the 1.8
[19:58:22] <skyfox00> (dispite having a buggy kernel and nvidia card)
[19:58:32] <skyfox00> on the 1.8, that is
[19:58:35] <SWPadnos> the kernel on the liveCD isn't buggy ...
[19:58:46] <SWPadnos> at least, not that we know of
[19:59:03] <skyfox00> the 1.8 had a buggy realtime kernel and the cvs-version of emc ran fine(exploded every once and awhile)
[19:59:28] <SWPadnos> did you try the liveCD also - just to do an exact comparison between the 1.8 and the 350?
[20:00:53] <skyfox00> yes, I just did a check with the same liveCD/halfile on both the 1.8 and the 350. 1.8 worked fine, 350 gave strange results...
[20:01:34] <SWPadnos> well, tedious though it ma ybe, I'd try every possible parallel port BIOS setting, with a liveCD reboot in between (possibly from power-down)
[20:01:58] <skyfox00> the liveCD on the 350 gave identical results to the cvs-emc2(compiled by me) testing I was doing yesterday.
[20:01:59] <SWPadnos> a couple of hours doing that will tell you if it's impossible to get it to work, or if it's just not configured correctly
[20:02:27] <SWPadnos> then you may be able to not boot from CD, but that's not as definitive a test
[20:03:44] <skyfox00> pluto MUST have the parport in EPP mode to work...
[20:04:22] <skyfox00> its an old computer with an updated BIOS so it may just not work right...
[20:04:27] <SWPadnos> EPP and ECP both support EPP. additionally, there are at least two versions of ECP to choose from
[20:04:45] <SWPadnos> or was that two versions of EPP? I don't remember
[20:05:01] <SWPadnos> the liveCD gives you a known working software setup with which to test the BIOS settings
[20:05:15] <skyfox00> it also take forever to boot ;)
[20:05:31] <SWPadnos> if you test BIOS settings with your own compile, then you don't know if it's the pluto software or the BIOS causing the problem
[20:05:36] <SWPadnos> yeah, that's why I say a coupld of hours ;)
[20:05:39] <SWPadnos> couple
[20:07:04] <skyfox00> I'm not ready for any gray hair yet.
[20:09:15] <skyfox00> besidesd, I would much prefer a faster computer anyway ;)
[20:09:25] <alex_joni> * alex_joni chose to switch directly to white hair
[20:09:32] <alex_joni> gray is not fun
[20:10:22] <skyfox00> * skyfox00 prefers to not beat a dead BIOS
[20:11:44] <skyfox00> well, I have stuff I need to do...(other than what I want to do) But I'm sure I'll be back with more questions sooner or later.
[20:15:03] <skyfox00> ok, I may be able to wrangle a 500Mhz for pluto... bye for now, though!
[20:15:25] <jepler> come back as long as it's with good news :-P
[20:15:27] <jepler> seriously .. good luck
[20:15:32] <jepler> I'm glad the hardware isn't damaged at least
[20:15:59] <skyfox00> yeah, me to, and thanks for having so much patience!
[20:17:08] <fenn> * fenn prefers a mad-scientist streak of white hair
[21:32:50] <alex_joni> good night all
[22:38:49] <jmkasunich_> jmkasunich_ is now known as jmkasunich
[22:49:12] <skyfox00> I just tested a 500Mhz (works with the pluto) with emc2, and got the go-ahead to turn it into a dedicated emc2 box!
[22:50:36] <jepler> phew!
[22:50:47] <skyfox00> ?
[22:51:06] <jepler> I mean, that's good news
[22:51:15] <skunkworks> jepler is happy it isn't his programming :)
[22:51:55] <jepler> welllll I still wouldn't rule that out
[22:51:56] <skyfox00> oh yeah, no kidding.... I'm happy his programming wasn't the problem to!
[22:52:00] <jepler> however it becomes an issue I can ignore
[22:52:05] <skunkworks> heh
[22:52:18] <skunkworks> skyfox00: what are you going to use the pluto to run?
[22:52:38] <skyfox00> you mean, what kind of machine?
[22:52:44] <jepler> there may be some detail of EPP programming that I overlooked
[22:52:56] <jepler> since "it works for me"
[22:53:05] <skunkworks> skyfox00: yes
[22:53:35] <skyfox00> the 500Mhz is an amd athlon, and could do EPP 1.x(dont remember) and EPP 1.9(I think)
[22:53:38] <jepler> for instance, I think I noticed that the drivers for elson's boards keep setting the SPP direction register when the EPP direction changes; pluto doesn't do that. Maybe if they did, it would have worked on skyfox's machine
[22:54:38] <skyfox00> skunkworks: dinky little sherline mill with the headstock spacer and lash adjustment nuts on x/y.
[22:54:39] <jepler> and I actually discovered I have a machine (1GHz fairly modern) where pluto doesn't work -- it behaves about like skyfox00 described his system
[22:55:02] <skunkworks> jepler: eww - yeck
[22:55:14] <skunkworks> skyfox00: cool
[22:55:20] <jepler> nothing wrong with dinky
[22:55:24] <jepler> my only milling machines are jokes
[22:56:00] <skunkworks> I have tried it on about 5 machines - all >1ghz without a problem.
[22:56:06] <jepler> well, "only milling machine". I have another CNC'd device but it's really a joke
[22:56:43] <skunkworks> jepler: no it isn't. I think you will get a lot of use out of it.
[22:56:52] <skyfox00> well, I want to(some day) designe/build small(model sized) diesel aircraft engines.... And then 'not dinky' helps alot...
[22:58:30] <skyfox00> http://jesseg.nikola.com:8080/foamboard/
[22:58:41] <skyfox00> I got to fly that beast ;)
[22:59:24] <jepler> fun hobby -- or I should say fun-sounding hobby
[22:59:52] <jepler> by the way, I think we have one user cutting foam with an XYUV machine and emc
[23:00:13] <skyfox00> fly-by-wire is fun.
[23:04:49] <skyfox00> what is an XYUV amchine?
[23:05:05] <jepler> XY UV describes the arrangement of the motors
[23:05:27] <jepler> X and Y you know
[23:05:31] <skyfox00> right
[23:05:38] <jepler> for a foam-cutting machine, U and V are parallel to X and Y but some distance away
[23:05:45] <jepler> 4 total motors
[23:06:09] <skyfox00> some sort of hot wire cutter?
[23:06:11] <jepler> in foam cutting you have a heated wire that runs between the two sides
[23:06:14] <jepler> right
[23:07:57] <skyfox00> it was a nighmare to cut the wing profile on that ping foam airplane... so I can see how a cnc foam cutter would be nice... such a machine was even talked about.
[23:10:04] <skunkworks> how did you cut it?
[23:10:10] <skyfox00> how it ended up being done, a stiff wire was bent into the profile and screwed down to a board, power was then applied and the 8 foot piece of foam was pushed through... perhaps its time to make a 8' x 4' x 6" cnc foam mill.
[23:11:51] <skyfox00> I think the farthest out we ever had it was about half a mile or so...
[23:12:09] <skyfox00> ugly thing aint it?
[23:12:24] <jepler> I don't know if you've heard of CNC Workshop, a yearly gathering in Illinois .. but one of the regulars has promised to bring a homemade CNC foam cutting machine next year
[23:13:07] <skunkworks> that works - I made some foam wings using a wire stretch between the chuck and tailstock of a 6ft lathe. metal paterns where stuck on each end of the foam block and 2 people (one on each end) pushed it thru. the wire was made hot by a beefy powersupply.
[23:14:43] <skyfox00> yeah, thats the redneck way to do it ;)
[23:14:56] <skunkworks> skyfox00: the video is awsome
[23:16:56] <skyfox00> hmmm, which one? we did a bunch and I can't remember which ones made it up on the net...
[23:22:58] <jepler> every time someone stumbles into the "net unhooked my existing signal" problem I second-guess the decision not to backport the code in TRUNK that makes it an error
[23:23:13] <skyfox00> lol
[23:24:44] <SWPadnos> hmmm. there are no min_output and max_output parameters for the 5i20 PWM
[23:25:19] <skyfox00> that could result in some strange behavior...
[23:26:47] <jepler> SWPadnos: there's code which is intended to clip to the max hardware limits
[23:27:00] <jepler> hal_m5i20.c: 914 // Truncate volts to DAC limits.
[23:27:43] <SWPadnos> ok, so it should be limited to _/- 10, doesn't the canonical DAC have limits though?
[23:27:42] <skyfox00> ok, I need to go grease my boots, cya leter!
[23:28:07] <jepler> see you skunkworks
[23:28:11] <jepler> SWPadnos: yes I believe it does
[23:28:17] <jepler> well .. not entirely sure
[23:28:21] <SWPadnos> ok - me either :)
[23:28:30] <jepler> I copied min-dc and max-dc from somewhere for pluto-servo
[23:41:29] <tomp> if there's no min or max, and i have +/- 2.5V device, do i use dac-nn-gain? and if it's not bipolar, do i use offset?
[23:43:37] <SWPadnos> tomp, yes, you use offset and scale to move the output to where you need it
[23:43:43] <SWPadnos> there should still be a limit though
[23:44:15] <tomp> i'm playing with it in halcmd right now...
[23:44:50] <SWPadnos> err - right, gain - not scale ;)
[23:46:44] <jepler> despites HAL trying to standardize things like analog outputs, I don't think that it specifies what voltage you get anywhere when you set gain=1, offset=0, value=1; for example, with pwmgen I think that gives you 100% duty cycle but which would give a variety of resulting output voltages depending on the output driver and filter used
[23:47:03] <jepler> it looks like the 5i20 driver is intended to give +1V on the 7i22 board in this case
[23:47:14] <SWPadnos> I thought that by default, inputting 1 should give you 1V on the output
[23:47:23] <SWPadnos> right
[23:47:24] <jepler> well, what is "the output"?
[23:47:48] <SWPadnos> sorry - the analog voltage output should be 1V for an input of "1" and gain of 1 and offset of 0
[23:47:51] <jepler> you can put other things than a 7i22 on the 5i20 pwm and direction pins
[23:48:11] <SWPadnos> I don't know if scale is supposed to be equivalent to gain or if they're reciprocals
[23:48:14] <jepler> maybe a single-ended 0-5 converter for instance
[23:48:23] <SWPadnos> sure
[23:49:32] <jepler> in pluto I think I make value -1 .. 1 be "full scale" with gain=1
[23:50:33] <jepler> since there's no particular DAC to go with it I dunno if that's a worse decision than anything else
[23:50:50] <SWPadnos> yeah - I'm not sure there's any possibility of uniformity
[23:50:51] <jepler> but maybe -3..3 should be full scale, since Voh is around 3v