#emc | Logs for 2009-07-13

Back
[01:05:19] <eric_unterhausen> bleah
[01:05:30] <Sordna> I agree
[01:06:12] <jepler> BigJohnT: 'git clean' removes files that are not versioned by git. read the manpage before using.
[09:22:41] <piasdom> g'mornin all
[09:25:24] <micges> good morning
[10:11:34] <EbiDK|AWAY> http://notalwaysright.com/shoot-first-worry-about-living-later/2208 Somone is working on their Darwin award..... :P
[12:50:15] <tom2> tom2 is now known as tom3p
[12:51:22] <tom3p> is there anything like css for the pyvcp xml files ( anything to reduce the repititious entries of same font color size... ) ?
[12:52:40] <SWPadnos> I don't think so
[12:52:46] <SWPadnos> feel free to add it though :)
[12:53:07] <tom3p> as a very accurate typist, i think it might be handy
[12:53:15] <SWPadnos> sure would
[12:53:15] <tom3p> ;)
[12:54:21] <archivist> a variable containing values, make a much easier to customise app
[12:54:53] <tom3p> something like #define even, will look further into it tonight
[12:55:11] <SWPadnos> I imagine a defaults section
[12:55:48] <SWPadnos> <defautls> / <led-color>RED</led-color> / <label-font>blahblah</label-font> ...
[12:55:50] <SWPadnos> err
[12:55:53] <SWPadnos> <defaults>
[12:56:07] <tom3p> i did find that rectled can be passed width & height , the rect allows nicer panel layout by forcing space use.
[14:04:18] <BJT-Work_> BJT-Work_ is now known as BJT-Work
[14:06:29] <f7ee> hello! Does anyone need list of packages to compile emc2-sim in kubuntu9.04?
[14:08:30] <SWPadnos> the wiki does ... :)
[14:16:44] <f7ee> s
[14:16:49] <f7ee> ok
[14:17:34] <f7ee> I can call it here in a minute
[14:20:32] <f7ee> "sudo apt-get install tk8.5-dev bwidget python2.5-dev python-imaging-tk libreadline-dev gettext libglib2.0-dev mesa-common-dev libgl1-mesa-dev libglu1-mesa-dev xorg-dev build-essential"
[14:21:01] <f7ee> you can check it from Kubuntu 9.04 LiveCD
[14:21:53] <f7ee> and "./configure --enable-run-in-place --enable-simulator && make clean && make"
[14:21:58] <SWPadnos> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?BasicSteps
[14:22:09] <SWPadnos> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2
[14:22:18] <SWPadnos> the BasicSteps page tells you how to edit the wiki
[14:22:36] <SWPadnos> this information should be put into section 5 of the Installing_EMC2 page :)
[14:22:44] <f7ee> Oh! I heard name of that guy, Shakespear, when I have written Makbet!!!
[14:22:58] <SWPadnos> o-tell-o
[14:23:12] <f7ee> OK, but not tonight. I'm saving this link
[14:23:26] <SWPadnos> ok. thanks for figuring it out (and documenting it :) )
[14:23:43] <SWPadnos> I think there's another step though - you have to remove tk8.4-*
[14:24:38] <SWPadnos> oh. and it's probably easier to get a checkout of the code if you install git
[14:25:03] <frallzor> yo
[14:25:11] <SWPadnos> yo-yo
[14:25:30] <frallzor> my first conversion is otw =D
[14:33:19] <tom3p> off the wall? on the water? on the way?
[14:33:57] <frallzor> on the way as allways ;)
[14:34:23] <tom3p> cool, mill? lathe? laser???
[14:34:48] <frallzor> converting a small proxxonmill
[14:35:31] <tom3p> nice, we dont have in USA but seen some pix
[14:35:31] <frallzor> but plans on making a portal-one in the future so I make everything futureproof =)
[14:35:59] <frallzor> yeah, they are pretty nice
[14:56:00] <jepler> I have this in my program:
[14:56:00] <jepler> M3 S200 (XXX: spindle speed) M8
[14:56:01] <jepler> G43 T4 M6
[14:56:12] <jepler> I'm surprised that emc waits for the spindle-at-speed indicator before moving to the toolchange position
[14:57:03] <cradek> me too, since that is a rapid
[15:01:21] <tom3p> does emc wait before other subsequent actions? like m3 s200 g0 x1?
[15:01:24] <tom3p> ´╗┐and maybe you're just testing combos, since you're asking for rpm just before it must be 0 for the xchg.
[15:01:43] <jepler> tom3p: it's a lathe
[15:02:17] <jepler> you can configure emc not to stop the spindle at toolchange, which makes sense in this case
[15:03:00] <tom3p> oh, spindle doesnt have the tool, gotcha
[16:24:38] <mozmck> SWPadnos: you don't have to remove tk8.4*, you just run configure this way...
[16:24:41] <mozmck> ./configure --with-tclConfig=/usr/lib/tcl8.5/tclConfig.sh --with-tkConfig=/usr/lib/tk8.5/tkConfig.sh
[16:25:16] <mozmck> I couldn't remove tk8.4 because I had other stuff that depended on it.
[16:37:20] <Jymmm> You're a glorified plumber in Alaska how do you get to your next job? By plane of course. And how long does a customer have to wait till you arrive? Up to six months.
[16:39:46] <jepler> joe the plumber is in alaska?
[16:40:30] <archivist> get fred instead
[16:40:46] <Jymmm> jepler: Right now on History Channel
[18:43:27] <frallzor> http://vimeo.com/5579091 progress =P
[18:48:03] <alex_joni> looks nice
[18:50:42] <frallzor> my damn cable is impossible to solder in the DB9-connectors, just wont stay where I want em =)
[18:50:53] <frallzor> but at least I could play a little
[18:51:38] <archivist_attic> tin connector and wire, bring together
[18:51:46] <frallzor> I am
[18:52:09] <frallzor> but think the copperwire are a bit too much for the pins
[18:52:16] <frallzor> *s
[19:05:14] <anonimasu> hm..
[19:05:21] <anonimasu> what kind of db9's do you have?
[19:05:23] <anonimasu> with solder cups?
[19:05:30] <frallzor> Id say so
[19:06:16] <frallzor> put some shrinktubing on the back of the wire to protect when done, but it failed too since the time i needed to get it to stick just a little made it shrink =)
[19:06:31] <anonimasu> I see, unless you have _huge_ wired that's odd :) it should be a problem
[19:07:04] <frallzor> well they arent huge, but it might be too huge for DB9s
[19:07:08] <archivist_attic> keep heatshrink up the wire well away, and dont take too long
[19:07:33] <frallzor> im thinking of just cutting have the copperwires away
[19:07:40] <frallzor> plenty of contact anyway
[19:07:47] <frallzor> *half the
[19:23:57] <Jymmm> frallzor: What are you using to solder, a plasma cutter?
[19:24:28] <frallzor> just a weller solderingstation, nothing fancy
[19:24:39] <Jymmm> got flux?
[19:24:42] <Jymmm> liquid
[19:25:12] <frallzor> nope, in the tin
[19:25:25] <Jymmm> Ah, *might* help if you can get some.
[19:25:38] <Jymmm> usually makes things go smoother
[19:25:50] <frallzor> but no matter how much flux I have im pretty sure these wires are a bit overdimensioned for the cups
[19:26:01] <frallzor> ill try cutting a few wires first
[19:26:02] <Jymmm> what size?
[19:27:17] <frallzor> 0.75mm2
[19:27:42] <Jymmm> what's that in imperial (AWG)?
[19:28:00] <frallzor> not a clue
[19:28:36] <frallzor> lets say 15?
[19:28:43] <Jymmm> 16
[19:29:01] <Jymmm> If so, sounds a tad too big, 18ga max typically
[19:30:03] <cradek_> cradek_ is now known as cradek
[19:30:15] <frallzor> half of the wires just bent back seems pretty ok
[19:37:18] <frallzor> never said this many bad words in a day before as Ive dont up til now
[19:51:11] <Guest579> Guest579 is now known as MikeGGG
[20:13:22] <MikeGGG> is there any reason not to set your servo_period equal to your base_period?
[20:13:55] <cradek> yes
[20:14:00] <SWPadnos> yes. the motion controller and PID generally run in the servo thread, and they use floating point
[20:14:22] <SWPadnos> they also take a lot more time to execute than the functions typically put in the base thread
[20:14:37] <cradek> of course your case may be different - ask a more specific question for a better answer
[20:15:24] <MikeGGG> well I just ran latency test. trying to figure out how to interpret those numbers into base/servo_period settings
[20:18:34] <MikeGGG> i guess the right answer is leave them alone until you have a reason to change them?
[20:19:07] <SWPadnos> it's probably more correct to set them up as you need them, and then leave them alone ;)
[20:19:55] <SWPadnos> how are you setting up your machine (stepconf, pncconf, by hand ...), and what kind of machine is it (stepper / servo, what hardware)?
[20:20:32] <MikeGGG> servo. m5I20.
[20:20:46] <MikeGGG> 20000 cpr encoders
[20:20:48] <SWPadnos> ok. in that case it's likely that you don't need a base thread at all
[20:20:50] <SWPadnos> cool
[20:20:59] <MikeGGG> hrm?
[20:21:12] <SWPadnos> the 5i20 does what would normally be done in the base thread: PWM/step generation and encoder counting
[20:21:23] <SWPadnos> there isn't much else that needs to be done that fast
[20:22:18] <MikeGGG> well i suppose that is good news. but now I am a little confused
[20:22:44] <SWPadnos> in what way?
[20:22:51] <MikeGGG> the PID loop remains in the real time portion right?
[20:22:56] <SWPadnos> yes
[20:22:56] <geo01005> better to be a little confused rather than alot.
[20:23:18] <SWPadnos> realtime can run multiple threads, and in fact you can only run HAL threads in realtime
[20:23:46] <MikeGGG> is the PID loop a HAL thread?
[20:24:05] <SWPadnos> PID is one of the things that gets done in the servo thread
[20:24:12] <SWPadnos> you could do it faster if you want
[20:24:35] <MikeGGG> ok
[20:25:01] <SWPadnos> have you read the HAL tutorial(s) yet?
[20:25:16] <MikeGGG> ...parts
[20:25:24] <SWPadnos> heh
[20:25:36] <frallzor> http://www.damencnc.com/tools/man/DCNC-IP54-1NM.pdf how many Amps would you recommend for this motor?
[20:25:37] <SWPadnos> well, you should read the part about threads and functions
[20:25:42] <MikeGGG> I don't actually have the hardware yet
[20:25:52] <Jymmm> too tor eeee al ?
[20:26:01] <SWPadnos> you can put whatever function you want in the thread that's most appropriate for your machine (within limits)
[20:28:29] <MikeGGG> could I put the PID and encoder read in the base thread?
[20:28:46] <SWPadnos> no, not really
[20:28:52] <MikeGGG> hm
[20:29:02] <SWPadnos> what's the purpose of making the PID and encoder reads go faster?
[20:29:23] <MikeGGG> it's a micromilling application
[20:29:34] <MikeGGG> so we want very fast accurate positioning.
[20:29:39] <SWPadnos> ok
[20:29:53] <SWPadnos> you can probably run the servo thread at 10 kHz or so with a 5i20
[20:30:31] <MikeGGG> really our concern at this point is that it is at least as fast as the National Instruments hardware we are using now
[20:30:32] <SWPadnos> are those 20000 cycle encoders or 20000 count encoders?
[20:30:42] <SWPadnos> which NI hardware?
[20:30:49] <MikeGGG> UMI-7764
[20:30:53] <SWPadnos> and what version of LabView?
[20:30:57] <SWPadnos> hmmm. I don't know that one
[20:30:58] <MikeGGG> er PCI 7764
[20:31:24] <MikeGGG> PCI-7354
[20:31:28] <MikeGGG> UMI-7764
[20:31:39] <MikeGGG> 20,000 count
[20:31:51] <MikeGGG> 5000 lines per rev. quadrature
[20:31:58] <SWPadnos> ok
[20:32:05] <SWPadnos> how fast is the PID loop on the NI board?
[20:32:11] <frallzor> hmm is it bad if I use like 0.3A over my psu? 1.8A for the A and it wont ever hardly be used together with the rest
[20:32:37] <SWPadnos> frallzor, you can probably do that intermittently without damage
[20:32:41] <MikeGGG> I think 93 us per axis
[20:32:45] <SWPadnos> if not, it'll just die on you
[20:32:53] <SWPadnos> execution time?
[20:33:02] <frallzor> SWPadnos its just for A, and itll not be used much so =)
[20:33:14] <MikeGGG> what does that mean?
[20:33:41] <SWPadnos> I'm asking if that's how long it takes to actually do the PID calcs (CPU execution time)
[20:33:50] <MikeGGG> i believe so
[20:33:58] <SWPadnos> so if you have 4 axes it would be 372 uS?
[20:34:07] <MikeGGG> correct
[20:34:18] <SWPadnos> how many axes do you run?
[20:34:20] <MikeGGG> 3
[20:34:32] <SWPadnos> ok. I think EMC2 will have no trouble beating that
[20:34:39] <MikeGGG> bitchin
[20:34:41] <SWPadnos> 279 uS is just under 4 kHz
[20:34:52] <SWPadnos> you should be able to do 5-10 with the 5i20
[20:35:16] <SWPadnos> and the overhead for adding more axes will be on the order of a few microseconds, not close to 100 like the NI unit
[20:35:24] <sed_> so we have completly rewired and isolated all signals on the machine and we are still having the problem with the stg limit fault when enableing amps when amps powerd.
[20:35:52] <SWPadnos> sed_, disconnect the limit switches and see if it keeps happening
[20:36:03] <SWPadnos> (don't disconnect e-stop, of course)
[20:36:41] <MikeGGG> to clarify, you mean 5-10 Khz base_period or is that 1/(5-10) KHz
[20:36:54] <sed_> so phisicly dissconect them or disable in hal?
[20:37:06] <SWPadnos> I mean you eliminate the base thread altogether, and run the servo thread at 5-10 kHz
[20:37:15] <SWPadnos> sed_, physically disconnect them
[20:37:23] <MikeGGG> ok, thanks!
[20:37:52] <SWPadnos> MikeGGG, I haven't tested the 5i20 at that speed, but I'm pretty sure you can get there
[20:38:05] <MikeGGG> well we shall see.
[20:38:17] <SWPadnos> 5 kHz no doubt (unless you have really weird kinematics), and probably 8-10 kHz
[20:38:27] <MikeGGG> just gotta make the case to the guys with the money.
[20:38:43] <MikeGGG> :)
[20:38:45] <SWPadnos> well, anything is less expensive than LabView and NI hardware
[20:39:04] <sed_> we are using normaly closed switches, dissconecting them give the fault as soon as we turn the software on now.
[20:39:07] <MikeGGG> i know. I might buy two, just in case I smoke one
[20:39:24] <SWPadnos> sed_, short them together at the input to the card
[20:39:33] <sed_> k hang on
[21:07:49] <jepler> I was recently running a 10kHz PID on 5i20. I think it was for 3 axes, though I only tested one axis.
[21:08:06] <jepler> (I might not have remembered to all all the functions to the 10kHz thread)
[21:10:17] <jepler> yes, I had added all 3 pids to the thread
[21:10:34] <jepler> I was only running pid at this rate, though, not the whole emc motion controller
[21:12:04] <jepler> (in the 1kHz thread, emc generates commanded position and the first pid generates a velocity command. in the 10kHz thread, the second pid takes a velocity command and generates a pwm command for a dumb servo amp -- and I should admit that it's not well-tuned yet, so I dunno if it's a win for my system)
[21:14:26] <sed_> ok that worked, so we are thinking that our limit switches share a cable with higher voltage amp limit switches is the issue
[21:14:37] <sed_> will the debounce hal command help???
[21:14:54] <MikeGGG> do you send the PWM command directly to the servo amp? I was looking at the 7I33 board which I think turns that into a true analog signal
[21:16:11] <jepler> MikeGGG: yes, it's pwm direct to an h-bridge, namely the 7i30
[21:16:37] <MikeGGG> ok
[21:18:03] <MikeGGG> the 5I20/7I33 have no opto-isolation. is that a big deal?
[21:18:49] <jepler> in my case I'm dealing with voltages no higher than 30V and I simply don't sweat it
[21:19:54] <MikeGGG> 30v is still sufficient to smoke a PC, no?
[21:20:00] <jepler> oh probably, but not people
[21:20:04] <MikeGGG> heh
[21:20:09] <MikeGGG> roger
[21:25:06] <jepler> when I blew up a stepper motor driver due to miswiring, it destroyed the driver with an internal short from Vmotor to power GND. It didn't damage any other components in the system.
[21:26:38] <cradek> 7i33 is very cheap compared to a smoked 5i20. it has real analog out and also has differential receivers for your encoders.
[21:27:00] <jepler> cradek: but no isolation, right?
[21:27:05] <jepler> that's where we started this discussion..
[21:27:15] <cradek> oh right, I guess not
[21:27:35] <jepler> you have 7i33, right?
[21:27:43] <jepler> do your servo amps have isolation at the inputs?
[21:27:45] <cradek> on mine, the logic supply on the servo amps is separate from anything else, so that ground gets hooked in
[21:28:01] <MikeGGG> well, hoping to shortly have
[21:28:07] <MikeGGG> not sure
[21:28:09] <cradek> I don't know if they are strictly isolated
[21:28:49] <MikeGGG> I was weighing some of the pros
[21:28:56] <MikeGGG> and cons of various hardware
[21:29:15] <MikeGGG> found a parport BOB that had opto isolation for ~$70
[21:29:42] <MikeGGG> just wondering why the mesa gear doesn't have it
[21:30:18] <MikeGGG> would like to know that the smoking will remain outside the PC
[21:30:23] <cradek> isolating analog and encoders (which must be very fast) is much harder than those digital outputs
[21:30:38] <MikeGGG> ahhh
[21:30:48] <cradek> but you'd have to ask mesa for a real answer
[21:32:37] <jepler> some servo amps are designed with isolation at their inputs, such as http://jelinux.pico-systems.com/pwmservo.html
[21:35:45] <MikeGGG> we have BMC-12H by motion controls group
[21:35:46] <MikeGGG> http://www.ametektip.com/index.php?option=com_content&view=article&id=200
[21:35:51] <MikeGGG> economical they are not
[21:39:17] <cradek> You must be registered and logged in to download Technical Documents.
[21:39:18] <cradek> Registration is free and instant. You may login immediately after completing the registration form.
[21:39:25] <cradek> fff
[21:39:44] <jepler> cradek: india:/usr/local/jepler/Desktop/BMC12H-installation-manual.pdf
[21:39:51] <jepler> (I signed up at that stupid site once)
[21:40:05] <tom3p> barnaby rubble, bedrock quarry.... fill it in ;)
[21:40:35] <cradek> hey look, it's late - I'm going home
[21:40:50] <MikeGGG> www.bugmenot.com
[21:40:51] <jepler> ooh so it is
[21:41:50] <jepler> bbl
[22:28:16] <Guest579> Guest579 is now known as MikeGG
[23:59:00] <Goslowjimbo> SWPadnos: Tried to check the SPI interface on Saturday, and had the oscilloscope die on me. 30 years and it up and quits.
[23:59:08] <SWPadnos> heh
[23:59:15] <SWPadnos> well, I guess SPI is not meant to be
[23:59:22] <SWPadnos> or you're meant to have a new scope or something :)
[23:59:27] <SWPadnos> dinnertime. bbl