#emc | Logs for 2008-11-16

Back
[00:45:16] <maddash> what's wrong with this: enum STATE {F=0,NP=1,NB=2,CS=3}; chargerstate STATE=FINE;
[00:45:24] <maddash> argh
[00:45:37] <maddash> I meant: enum charger {F=0,NP=1,NB=2,CS=3}; chargerstate STATE=FINE;
[00:46:09] <maddash> gcc keeps throwing me, "syntax error before STATE; data definition has no type or storage class"
[00:57:15] <fenn> is that supposed to be C?
[00:57:31] <maddash> YES
[00:57:34] <maddash> yes*
[00:58:53] <fenn> i suppose the second line should be something like: charger state=F;
[00:59:05] <fenn> charger STATE=F;
[00:59:12] <fenn> blah
[00:59:21] <fenn> either way
[02:51:06] <jmkasunich> http://media.armadilloaerospace.com/2008_10_24/LLC_Level_1_flight_2.mpg
[02:51:36] <jmkasunich> John Carmack (author of Doom) has been busy
[03:20:35] <jepler> sure is a realistic-looking lunar lander simulation simulation
[03:31:43] <GNieport_> hi everyone. I am looking at controlling a dot peen marking head with EMC, such as http://www.dapramarking.com/dot-peen-marking.htm I would like to ask for opinions on how to control the "Z" or pin strike.
[03:32:39] <GNieport_> The pin would ultimately be controlled by a digital output driving a solenoid valve.
[03:33:23] <GNieport_> How would I trigger a signal when Z is commanded down, for example?
[03:34:28] <SWPadnos> use a comparator in HAL, and have the output go to the solenoid
[03:34:43] <GNieport_> Perhaps set EMC to only run the X and Y axes, but have a G code command sent to Classic Ladder for each strike...
[03:34:46] <jepler> http://cscott.net/Projects/FabClass/freedom/
[03:34:55] <GNieport_> hmm
[03:35:18] <GNieport_> SWPadnos, comparing position register?
[03:35:25] <SWPadnos> yes
[03:35:54] <SWPadnos> actually a window comparator - wcomp
[03:36:01] <GNieport_> Good, any concerns with cycling Z 50-100 times per second?
[03:36:05] <SWPadnos> it has over, under, and "in the range" output bits
[03:36:14] <GNieport_> Excellent
[03:36:37] <SWPadnos> as long as the 1kHz cycle is fast enough, or you can run the servo period faster, shouldn't be any issues
[03:37:16] <jmkasunich> jepler: very realistic ;-)
[03:37:37] <SWPadnos> the wcomp also gives you some hysteresis, so you can set the "o" trip to -0.1 and the off trip to 0.1 - that can give you a dwell time
[03:37:47] <jmkasunich> I've been following those guys a couple of years now
[03:37:57] <SWPadnos> that was a nice landing
[03:38:10] <SWPadnos> and much harder than on the moon for several reasons
[03:38:13] <jepler> other routes include: the "digital output" code, the M1xx custom gcode, spindle control codes, and many others
[03:38:46] <jepler> an m1xx custom code can let you set the solenoid bit, delay, clear it, and delay again
[03:38:58] <jepler> there was discussion about this on the mailing list just today (about collet open/close though)
[03:39:03] <GNieport_> Great, several ways to do it!
[03:39:16] <SWPadnos> that should work, but for 100Hz peening, I don't know that I'd rely on userspace
[03:39:30] <SWPadnos> 100Hz peeing should be fine though :)
[03:40:04] <jepler> true
[03:40:27] <jmkasunich> GNieport_: does the on-time of the solenoid need to be varied? is it critical?
[03:40:57] <GNieport_> The factory controller used to be pretty zippy. About twenty 12 point chars per second. I would be pleased with one CPS.
[03:41:18] <GNieport_> jmkasunich, not sure yet
[03:41:33] <jmkasunich> do you need to position, stop, hit, position, stop, hit? or just move while hammering?
[03:42:16] <GNieport_> My calibrated eyes tell me it was hammering while in motion :)
[03:42:17] <jmkasunich> if the latter you could probably just do a pulse generator thing in HAL, optionally gated on and off by g-code or Z position
[03:42:45] <SWPadnos> yeah, it depends on whether you consider the work to be a dot-matrix image, or a dotted line
[03:43:06] <GNieport_> The pin has a spring return, like a doorbell. One air pulse to extend, mark, and retract
[03:43:35] <SWPadnos> the dotted line would require a velocity-based frequency generator to get consistent dot spacing
[03:44:18] <GNieport_> The function is more dot-matrix...
[03:44:46] <jmkasunich> yeah, just saw the website says 5x7 or 7x9 matrix
[03:44:56] <GNieport_> My initial understanding is that the marking head is looking for a quick voltage to trigger the air.
[03:45:11] <jmkasunich> although they also say that "continuous line" fonts are available
[03:45:26] <GNieport_> The steppers move the marking stylus into each dot location
[03:45:45] <jmkasunich> the line thing would be easy - just buzz the stylus and move it around using g-code
[03:45:50] <jmkasunich> like tattooing ;-)
[03:45:54] <GNieport_> :P
[03:46:29] <jmkasunich> it is going to take either reverse engineering or some trial and error to determine the proper pulse length I bet
[03:46:29] <GNieport_> This particular marker is about 20 years old now :)
[03:46:40] <GNieport_> I'm sure :)
[03:47:33] <jmkasunich> there are several ways to tackle the problem in EMC, you really need more info to decide which is best
[03:48:06] <jmkasunich> if the pulses are several mS long and can tolerate length granuarlty of a mS (or maybe 1/2mS) then it is quite easy
[03:48:37] <jmkasunich> I think EMC has a one-shot component that could generate the pulse
[03:49:01] <SWPadnos> there's timedelay, though it uses FP
[03:49:13] <jmkasunich> if you need sub-mS pulses, then you might have to write a component, but if you know basic C it still doesn't seem that tough
[03:49:16] <SWPadnos> and a one-shot
[03:49:23] <GNieport_> Ah, yes, I haven't yet dug the machine out of the junque pile
[03:49:31] <SWPadnos> oneshot also uses FP
[03:49:39] <SWPadnos> (probably not a problem, but good to note)
[03:50:38] <GNieport_> jmkasunich, I doubt the pulses are <mS
[03:50:58] <SWPadnos> if it goes 100Hz, they may be pretty short (probably not <1ms though)
[03:51:11] <jmkasunich> it might be 1mS on, then 10mS for recovery
[03:51:23] <GNieport_> the 100 Hz is a guesstimate
[03:51:25] <jmkasunich> I bet strike time is quicker than vent/retract time
[03:52:15] <GNieport_> Indeed
[03:52:48] <ds3> hmmm
[03:52:52] <SWPadnos> roughly 20 CPS is a better number to use. there are 35 or 63 dot locations for each character, and if 15% of them are on, you're looking at ~100-200 strikes/sec
[03:53:05] <jmkasunich> looks like the current models from Dapra are all-electric
[03:53:05] <SWPadnos> (if either of our estimates are close ;) )
[03:53:31] <ds3> EMC controlled "type writer"?
[03:53:47] <SWPadnos> peening poker
[03:54:40] <ds3> ultimately, it'll be like a printer, right?
[03:55:02] <fenn> no way your solenoid valve is going to do 100Hz
[03:55:03] <jmkasunich> GNieport_: you are aiming at the same basic function that the original machine had, right? text only, fixed dot matrix? (as opposed to drawing arbitrary patterns)
[03:55:43] <ds3> or is this a CNC version of shot peening for workhardening materials?
[03:55:43] <jmkasunich> fenn: there are fast valves, and I assume GNieport_ is planning to use the valve that is in the machine now, not some random hardware store valve
[03:55:52] <jmkasunich> ds3: printing
[03:56:28] <jmkasunich> hmm - the current model is described as "Highly Productive: Marks several characters per second."
[03:56:36] <jmkasunich> "several" < 20
[03:56:37] <fenn> wow
[03:56:47] <fenn> still fast
[03:56:58] <fenn> only one pin?
[03:57:06] <jmkasunich> far as I can tell, yes
[03:57:39] <jmkasunich> http://www.dapramarking.com/dotpeen/modular-vids.htm
[03:57:43] <jmkasunich> vid of it in action
[03:59:11] <ds3> there is a "edm-ish" version of it described in the current issue of digital machinist
[04:00:18] <GNieport_> ds3 Marking characters as a first tier, logo later if feasible
[04:00:31] <GNieport_> er, jmkasunich :)
[04:01:01] <jmkasunich> the vids seem to show constant or near constant pulse rate
[04:01:16] <SWPadnos> yeah, in the 50-ish Hz range maybe
[04:01:23] <GNieport_> These newer machines lay down a finer dot pitch
[04:01:25] <SWPadnos> 50-80, my calibrated ears say
[04:01:31] <GNieport_> lol
[04:01:37] <SWPadnos> (though I'm using an uncalibrated internet connection ;) )
[04:01:53] <jmkasunich> varies - the vid that is writing on a round steel bar is quite a bit slower than that
[04:02:00] <ds3> SWPadnos: ears traceable to NIST? ;)
[04:02:13] <SWPadnos> yeah, they've been there :)
[04:02:20] <SWPadnos> there are probably photos to prove it
[04:03:01] <jmkasunich> GNieport_: I think my first approach would be to cook up a pulse generator in HAL, and gate it with the output of a comparator looking at Z
[04:03:01] <GNieport_> Hmm, that Multi 4 handheld gun is slower than this unit, but the second marking tool shown is about the same
[04:03:20] <GNieport_> Mmm, cooking with HAL
[04:03:50] <GNieport_> That sounds a fine approach
[04:04:12] <ds3> slightly different topic - has anyone tried "surfacing" with just a drill bit and plunging (inplace of the lateral movements with a ball EM)
[04:04:35] <GNieport_> I appreciate all of the input
[04:05:04] <jmkasunich> two one-shots could probably do it - the first one would drive the valve and set the on-time, the second would trigger on the end of the first one, and would time the off-time
[04:05:14] <jmkasunich> the 2nd output would loop around and drive the first again
[04:05:24] <jmkasunich> although getting it started might be a trick
[04:06:50] <jmkasunich> another approach would be to run the sawtooth output of siggen into a comparator
[04:07:12] <jmkasunich> siggen.frequency sets the freq, and the value on the other comparator output sets the duty cycle
[04:10:09] <SWPadnos> no reason you couldn't use a square wave driving the one-shot
[04:10:33] <jmkasunich> yeah, except that you'd still need a comp, since the square wave is a float and the oneshot wants a bit
[04:10:37] <GNieport_> That approach would be used for a continuous dot stream, right?
[04:10:39] <SWPadnos> true
[04:10:48] <SWPadnos> GNieport_, yes
[04:10:59] <jmkasunich> continuous, or gated by Z depth
[04:11:23] <jmkasunich> if you want to move, hit, move, hit, move, hit, I bet the speed will suffer
[04:11:33] <GNieport_> ah, travel to beginning of character, drop Z, move at constant rate while dots are laid...
[04:11:38] <jmkasunich> yep
[04:11:49] <SWPadnos> yeah - I'm not sure how blending would affect the Z-as-driver method
[04:12:13] <GNieport_> Could use exact stop
[04:12:21] <GNieport_> but, slow, as you point out
[04:12:27] <SWPadnos> you'd have Z moves though
[04:12:43] <jmkasunich> you probably want blending on the XY moves, for speed
[04:12:59] <SWPadnos> actually, add Z to a triangle wave, then use a wcomp on that
[04:13:18] <jmkasunich> or compare Z to the triangle (or sawtooth)
[04:13:34] <SWPadnos> sure, that would set the width
[04:13:40] <SWPadnos> pulse width
[04:14:27] <jmkasunich> Z blending can be made negligable, zince "up" and "down" depths are arbitrary
[04:14:29] <GNieport_> Great, text should be doable
[04:14:37] <jmkasunich> you could make up be 0.001" and down be 0.000"
[04:14:55] <GNieport_> Exatly
[04:14:59] <SWPadnos> one potential issue is that you're not guaranteed to get the identical pulse stream, since motion isn't synchronized to the siggen
[04:15:29] <SWPadnos> unless you use feedhold or something at the start of each dot run
[04:15:53] <jmkasunich> it wouldn't be hard to write a gated pulse generator, instead of gating a free-running generator
[04:16:04] <SWPadnos> true
[04:16:13] <GNieport_> Well, if that turns out to be an issue, the M codes would come in handy
[04:16:16] <SWPadnos> you could even make one with a bit output :)
[04:16:32] <GNieport_> Really, it is a two-axis system in my mind
[04:16:48] <SWPadnos> yep, it is
[04:16:50] <jmkasunich> the problem with M1xx codes is that they are handled in user space, which is dog slow at the speeds we are discussing
[04:17:03] <GNieport_> Ah, I forgot your note
[04:17:16] <SWPadnos> and those should require exact stop (whether you use G64 or not)
[04:17:32] <SWPadnos> and they're susceptible to user-space timing issues
[04:18:06] <GNieport_> Has "pen-down" for a plotter been done before?
[04:18:23] <GNieport_> mIght be applicable in part
[04:18:26] <SWPadnos> I know it's been discussed, I don't know for sure if someone did it
[04:18:47] <SWPadnos> hmmm. actually, that Jack guy in Peoria - that was a CNC punch - I'm not sure what he did
[04:18:59] <GNieport_> Okay, so M1XX isn't feasible at speed.
[04:19:50] <jmkasunich> SWPadnos: that was a lot slower
[04:19:59] <jmkasunich> the machine didn't actually position the work
[04:20:08] <SWPadnos> yeah, that's true
[04:20:14] <jmkasunich> it positioned stops, then a human slid the work against the stops, and hit a foot pedal IIRC
[04:20:28] <SWPadnos> oh right, it was a manually-operated punch
[04:20:41] <SWPadnos> well GNieport_, you'll just have to have a really fast foot :)
[04:20:49] <GNieport_> Yaya
[04:24:37] <GNieport_> EMC shouldn't choke on 0.001 Z rapids at the marking speed.
[04:24:53] <GNieport_> Set the Z accel very high.
[04:25:01] <SWPadnos> hmmm
[04:25:03] <SWPadnos> yes and no
[04:25:11] <jmkasunich> best way to find out is to test it
[04:25:18] <GNieport_> really? darn :)
[04:25:26] <jmkasunich> program some XY moves, then a short Z, then some more XY, etc
[04:25:29] <GNieport_> and write a Wiki
[04:25:37] <jmkasunich> then use halscope to observe the position commands
[04:25:44] <SWPadnos> the block rate should be no problem, but blending may cause Z to cross the threshold at spots where you don't expect it
[04:26:03] <jmkasunich> (you don't need any hardware to do that - you can test before you build anything else)
[04:26:17] <GNieport_> g64.whatever to set the proximity to programmed path
[04:27:07] <GNieport_> jmkasunich, I see what you mean
[04:27:29] <SWPadnos> sim is handy that way :)
[04:27:42] <GNieport_> Blech, I wish the garage wasn't cold and rainy
[04:27:50] <SWPadnos> sim doesn't need RT
[04:28:31] <GNieport_> This is going to be a leap for me, I have to figure out the stepper stylus driver too. I've only set up a servo machine before now :)
[04:28:44] <GNieport_> wah.
[04:28:56] <GNieport_> Thanks again, guys
[04:29:18] <SWPadnos> sure, have fun with it
[04:29:36] <GNieport_> Yep, I'll put up some docs when I have something to report
[04:30:01] <GNieport_> That'll be about Fall 2010
[04:30:56] <GNieport_> I still have to add a servoed Z axis to my knee mill
[04:33:57] <GNieport_> Just finished this gage despite the lack of Z
[04:33:58] <GNieport_> http://imagebin.ca/view/9KOTNh.html
[04:34:21] <jmkasunich> neat
[04:34:26] <jmkasunich> what's it for?
[04:34:42] <GNieport_> To engrave the ID number, I put in M0 everywhere I had to crank the knee up and down :)
[04:35:14] <GNieport_> measuring a part length, from the oval cutout to the notched black oxide column
[04:36:30] <GNieport_> That was my first dollar earned with EMC. I toast all of you!
[04:38:10] <GNieport_> I did the roughing with a jogwheel pulsegen, and used 2 axis for the arcs
[04:42:09] <jmkasunich> does the mill have a quill?
[04:42:38] <GNieport_> Yes, I have a ball screw that I picked up to mount on the front of the quill
[04:43:13] <jmkasunich> I was confused when you wrote that you were moving the knee for the text
[04:43:27] <GNieport_> The quill has no stops
[04:43:31] <jmkasunich> but I guess if you need repeatability on a manual, the knee is better
[04:43:42] <GNieport_> I had to control the depth, yes
[04:44:08] <GNieport_> The traditional stop rod is completely gone, the bosses ground away by prior owner
[04:45:04] <GNieport_> Thankfully, the tapped holes in the quill ram are in great shape
[05:06:43] <fenn> GNieport_: you should just use a cursive font :)
[05:08:10] <GNieport_> fenn, I'll give that a try
[08:27:49] <Memphis> Memphis is now known as Gamma-x
[08:28:07] <Gamma-x> whats up everyone
[08:38:05] <OoBIGeye> breakfast
[08:38:43] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[12:59:39] <BigJohnT> hi alex_mobile
[15:23:20] <fragalot> hmm
[15:23:35] <fragalot> my laptop keeps crashing for no reason, so i just opened it to see if it's not just overheating because of dust
[15:23:47] <fragalot> and... It's as clean inside as if it just came off the assembly line
[15:24:13] <fragalot> there is however, a complete lack of a heatsink worth mentioning,.. But i think it makes up by using the entire chassis as a heatsink, since the chassis is completely made out of copper
[15:44:21] <alex_mobile> Hi BigJohnT
[15:44:37] <BigJohnT> hi alex_mobile
[15:45:13] <alex_mobile> Slow day?
[15:46:07] <BigJohnT> not slow enough :)
[15:46:57] <BigJohnT> working on a 3-D deer in the woods backlit by a candle thingy for the wife
[15:47:09] <alex_mobile> Heh. It's pretty slow here. No PC access all day.
[15:47:26] <BigJohnT> and learning docbook xml xsl css etc
[15:47:40] <BigJohnT> you out and about today
[15:47:56] <alex_mobile> Like the german bogen?
[15:49:08] <alex_mobile> Yeah. 2 birthdays i need to attend.
[15:49:26] <BigJohnT> having fun then :)
[16:15:39] <Poincare> if I want to drill holes in a pre-developed pcb card, is there a way to compensate for misalignment (rotate the drill file) ?
[16:17:35] <alex_mobile> Rotate the pcb
[16:24:36] <Poincare> alex_mobile: that's a lot of trial and error :-) i have used a system a long time ago where we had to touch of 2 know points on the board, the routing software then compensated for the misalignment
[16:24:54] <Poincare> was wondering if there's something like that available with emc
[16:25:48] <StefanS__> hi guys
[16:27:46] <StefanS__> could someone help me in configuring the mesa 5i20 card with emc 2.2.7 ?
[16:29:06] <alex_mobile> Usually it's best to ask specific questions to get good answers.
[16:29:09] <fragalot> Poincare: you could do that manually
[16:29:35] <fragalot> Poincare: just touch one point, make sure it doesn't move, go to another, touch that so it's aligned, and lock the pcb in place..
[16:30:08] <StefanS__> :-)
[16:30:11] <JymmmEMC> fragalot: I don't think that'll compensate for any skew though.
[16:30:36] <JymmmEMC> maybe it would, I just dont know
[16:30:44] <jepler> Poincare: no, emc doesn't presently do coordinate system rotation; it would be a welcome addition if someone codes it, though
[16:31:14] <JymmmEMC> * JymmmEMC notes.... That's jepler's standard response btw =)
[16:31:32] <anonimasu> coordinate system rotation is really useful btw..
[16:31:33] <anonimasu> :p
[16:31:59] <JymmmEMC> jepler: Hey, anonimasu JUST volunteered to write it!
[16:32:11] <fragalot> lol
[16:32:30] <anonimasu> the control on my big mill already does it ;)
[16:32:52] <jepler> long ago I wrote "ckins", a kinematics module that adds a fake C axis to a traditional XYZ machine. It would need to be updated to run with current versions of emc: http://emergent.unpy.net/files/sandbox/ckins.c
[16:33:04] <JymmmEMC> anonimasu: There ya go... you already have a test bed to use to debug with
[16:33:24] <anonimasu> lol
[16:33:29] <cradek> if you can't write it, making a spec that describes how it works, including how it interacts with the various kinds of offsets and coordinate systems, for all to comment on, would be the next best thing
[16:33:32] <jepler> you'd also need a procedure to get the C-axis rotation from the two points
[16:33:56] <anonimasu> hm, it works like this, you rotate everything around the point where you are at..
[16:34:03] <StefanS__> emc stops with something like "hm2/hm2_5i20.0: invalid cookie, got 0x00000100, expected 0x55AACAFE"
[16:34:20] <anonimasu> coordinate systems and stuff too
[16:34:21] <JymmmEMC> Can you actually determin skew from two points?
[16:34:28] <anonimasu> well, except for the ref system..
[16:34:30] <seb_kuzminsky> StefanS__: that means your firmware didnt load correctly
[16:34:43] <cradek> anonimasu: right but what happens to everything else? if you rotate g54 what happens when you switch to g55? or if you apply a g92 while rotated?
[16:34:48] <seb_kuzminsky> we can help you but we're going to need a bunch of information from you
[16:34:48] <StefanS__> yep, and why ?
[16:35:00] <jepler> mailing list thread about a similar problem: http://www.mail-archive.com/emc-users@lists.sourceforge.net/msg09589.html
[16:35:06] <jepler> (invalid cookie)
[16:35:19] <jepler> (though that was on a different mesa board)
[16:35:32] <anonimasu> cradek: it rotates the coordinate system around the new point
[16:35:47] <seb_kuzminsky> jepler: yeah i think the 5i20 isnt susceptible to the kinds of io errors that caused that problem (it was on an epp board)
[16:36:14] <seb_kuzminsky> StefanS__: please put your hal files and the output of "dmesg" on pastebin.ca and tell us the link here
[16:36:39] <anonimasu> jepler: I didnt try that, I can do it someday if I can remember
[16:36:54] <seb_kuzminsky> StefanS__: most likely you didnt give a "firmware=" part to your hm2_pci config modparam
[16:37:39] <StefanS__> uhh, you're fast
[16:38:14] <seb_kuzminsky> you should see me after another cup of coffee ;-)
[16:38:50] <StefanS__> my hal file just tries to load the rt-module by "loadrt hm2_pci config="firmware=hm2/5i20/SVST8_4.BIT num_encoders=3 num_pwmgens=3 num_stepgens=3"
[16:39:11] <seb_kuzminsky> well that looks good
[16:39:33] <seb_kuzminsky> what does "md5sum /lib/firmware/hm2/5i20/SVSST8_4.BIT" say?
[16:39:45] <StefanS__> right now there is nothing else in the hal file, could that be a problem ?
[16:40:14] <seb_kuzminsky> StefanS__: there must be a "loadrt hostmot2" before that, right?
[16:40:28] <anonimasu> jepler: reading the document, it seems that you should always program datum shifting before other operations
[16:40:34] <anonimasu> jepler: scaling/mirroring/rotaiton
[16:40:36] <anonimasu> rotation
[16:40:48] <jepler> StefanS__: did you use the board with an earlier version of emc2, or is 2.2.7 your first try?
[16:41:54] <StefanS__> it's the first try, but the original m5i20 examples where able to load the firmware
[16:42:10] <anonimasu> jepler: it means that you have to cancel rotation then change to a new system and then rotate it again
[16:42:23] <seb_kuzminsky> StefanS__: m5i20 is a totally different driver that is way more mature and way easier to configure :-/
[16:42:50] <seb_kuzminsky> StefanS__: what's the md5sum of the BIT file and the dmesg output please? Without that we're shooting in the dark here
[16:42:52] <jepler> if hal_m5i20 works then it points to software/configuration problems rather than hardware problems, which is nice to know
[16:43:12] <StefanS__> md5 says dcf2cabf2a4b3fc08d8b783bbaaddeab
[16:44:14] <seb_kuzminsky> that't the rigth md5sum, so the firmware file exists and is correct (not corrupted), so that's good
[16:44:35] <seb_kuzminsky> I like how part of the md5sum says "bbaadd" - it's superbad
[16:45:03] <jepler> hah. "The take home message is, when you're dealing with something you're maybe not 100% familiar with, and it's performance is a bit off, cranking the dial from 3.5 to 10 first thing is probably not a good approach."
[16:45:10] <SWPadnos> and if you squint while you say it, bbaaddeab could almost rhyme with Muad'dib
[16:45:25] <seb_kuzminsky> "he is the cuisinart hatrack!"
[16:45:35] <SWPadnos> the wizard's hatrack
[16:46:52] <StefanS__> here's the pastebin link http://pastebin.com/m64c2fa18
[16:47:48] <jepler> #
[16:47:49] <jepler> [18443.674275] hm2_pci: Unknown symbol hm2_register
[16:48:05] <jepler> aha, so the problem may be the missing loadrt for the core hm2 component, what's-its-name
[16:48:10] <seb_kuzminsky> thanks StefanS__
[16:48:36] <seb_kuzminsky> StefanS__: wait, that's a different error than what you reported to us
[16:49:03] <seb_kuzminsky> jepler is right, you're missing the hostmot2 driver, say "loadrt hostmot2" before "loadrt hm2_pci ..."
[16:49:26] <StefanS__> uhhh, just a moment...
[16:49:48] <seb_kuzminsky> brb
[16:51:57] <StefanS__> damn, it seems to work better. which document did i miss to read :-)
[16:52:03] <seb_kuzminsky> lol
[16:52:17] <seb_kuzminsky> did you read the hm2_pci and hostmot2 manpages?
[16:52:39] <seb_kuzminsky> they're not exactly how-to guides, but i think they contain the information you need
[16:52:46] <seb_kuzminsky> if not they need to be updated
[16:54:37] <StefanS__> yes, i tried them, but they are hard stuff when digging deeper the first time
[16:54:42] <BigJohnT> seb_kuzminsky: I see we have a section on the 5i20 in the Hardware chapter of the Integrators manual
[16:54:53] <seb_kuzminsky> oohh! yes please!!
[16:54:57] <BigJohnT> is it up to date?
[16:55:13] <seb_kuzminsky> i think it's all about the old m5i20/hostmot-4 driver
[16:55:17] <seb_kuzminsky> there's nothing about hostmot2 in there yet
[16:55:39] <BigJohnT> if you can forward me the info I can add it
[16:55:49] <seb_kuzminsky> BigJohnT: great! thanks!
[16:55:54] <BigJohnT> np
[16:56:00] <StefanS__> btw, i was invited by alex joni on a german cnc forum to create an example configuration for the new driver
[16:56:05] <seb_kuzminsky> BigJohnT: your documentation-fu is strong
[16:56:11] <seb_kuzminsky> StefanS__: that would be most welcome
[16:56:28] <seb_kuzminsky> StefanS__: there's a sample configuration for 3-axis servo control with hostmot2 that unfortunately didnt make it into 2.2.7
[16:56:28] <BigJohnT> :)
[16:57:03] <seb_kuzminsky> StefanS__: you can see it here: <http://cvs.linuxcnc.org/cvs/emc2/configs/hm2-servo/<
[16:57:07] <seb_kuzminsky> er, <http://cvs.linuxcnc.org/cvs/emc2/configs/hm2-servo/>
[16:57:38] <seb_kuzminsky> it uses the 7i43 instead of the 5i20 like you have, but a simple global search-and-replace should take care of that
[16:57:56] <seb_kuzminsky> (and change which driver it loads, and which firmware it sends)
[16:58:06] <seb_kuzminsky> StefanS__: so you're up and running now? all good?
[16:59:22] <StefanS__> firmware downloading works
[16:59:27] <seb_kuzminsky> yay!
[16:59:38] <seb_kuzminsky> are you doing all this using halcmd?
[16:59:52] <alex_mobile> StefanS: glad to see you're working on it.
[17:00:01] <seb_kuzminsky> if so, run "show pins" and you should see all the 5i20 stuff (there's a ton)
[17:00:08] <StefanS__> now i have to assign pins and functionality
[17:00:30] <seb_kuzminsky> this is where the hm2-servo config i linked to you might help
[17:00:49] <seb_kuzminsky> ok, i'm off to have a shower - i smells funky
[17:00:51] <seb_kuzminsky> bbl folks
[17:01:14] <StefanS__> unfortunately i'm not familiar with the hal stuff, so i have to learn a lot
[17:01:24] <alex_mobile> TMI
[17:02:03] <alex_mobile> * alex_mobile is really slow today
[17:03:08] <BigJohnT> StefanS__: in AXIS sometimes it is easier to do Machine/Show HAL Configuration to see the pins and signals
[17:04:32] <StefanS__> just had a look there..great!!
[17:05:09] <BigJohnT> in the watch window you can see the pins change state etc
[17:08:01] <BigJohnT> StefanS__: take a look at this page http://www.linuxcnc.org/docview/html//hal_tutorial.html#r1_5_2
[17:08:12] <fragalot> y'know, i'd like it if you could edit the G-code inside axis, instead of having to open another editor, modify, reload, etc..
[17:19:29] <nots> nots is now known as nots_
[17:19:35] <nots_> nots_ is now known as notsu
[17:24:18] <JymmmEMC> well, that was weird
[17:24:57] <SWPadnos> http://www.newegg.com/Product/Product.aspx?Item=N82E16822145167
[17:25:11] <SWPadnos> damn. $90 after rebate (on up to 5)
[17:25:26] <SWPadnos> and even less if you buy cables with it, like $3-4 less (in qty 5)
[17:25:33] <JymmmEMC> whats wrong with the WD green ones?
[17:25:41] <SWPadnos> nothing
[17:25:50] <SWPadnos> they may be a little slower, but they do take less power
[17:25:58] <JymmmEMC> oh what cpu do you have for your new system?
[17:26:16] <SWPadnos> the server with the TB drives in it, or the workstation with the fried motherboard?
[17:26:26] <JymmmEMC> fried
[17:26:34] <SWPadnos> AMD Phenom 9850 X4
[17:26:50] <JymmmEMC> which is what a quad core?
[17:26:50] <SWPadnos> or Phenom X4 9850
[17:26:59] <SWPadnos> yes
[17:27:02] <SWPadnos> 2.5 GHz
[17:27:13] <JymmmEMC> so nothing really special
[17:27:15] <JymmmEMC> ?
[17:27:24] <SWPadnos> does a full EMC2 build (sim) in about 20 seconds
[17:27:29] <JymmmEMC> (no sarcasim intended)
[17:27:34] <SWPadnos> including deps
[17:27:53] <SWPadnos> nope, second from the top of the line in AMD-land, not as fast as the $$$$$ Intels
[17:28:02] <JymmmEMC> ok, so it's a very nice cpu, but nothing special like the Xeons are/used to be
[17:28:06] <SWPadnos> ok, time to run outside now :)
[17:28:08] <SWPadnos> no
[17:28:13] <JymmmEMC> k
[17:28:19] <SWPadnos> not multi-socket capable or anything
[17:28:28] <JymmmEMC> k
[17:33:20] <SkullWorks_PGAB> OK I got in here just a little too late... Looks like Seb is already off to the showers.
[17:38:05] <alex_mobile> He's not Kramer. Can't stay in there forever.
[17:40:52] <JymmmEMC> if I actually liked seinfeld, I might find that funny.
[17:42:52] <SkullWorks_PGAB> Seinfeld had to be the most irritating show since three's company.
[17:44:14] <SkullWorks_PGAB> well maybe not - I'm not a good judge - I haven't turned on my TV more that twice in 2008
[18:07:39] <Gamma-x> hey everyone!
[18:07:54] <Gamma-x> BigJohnT, whats up brother.
[18:08:30] <BigJohnT> trying to figure out the syntax of a makefile :)
[18:08:39] <BigJohnT> how about u?
[18:09:11] <Gamma-x> uhhhm got a new job, moved to dc and uhhh hating life, thinkin about joinin army reserves for sumtin exciting haha
[18:09:31] <BigJohnT> that sounds like fun
[18:09:39] <BigJohnT> what ever happened to the mill?
[18:09:53] <Gamma-x> still have it
[18:09:57] <Gamma-x> havnt touched it though
[18:10:01] <Gamma-x> the pid thing was killing me.
[18:10:30] <Gamma-x> i could get the mill to move but without the rotary scales or w/e it just couldnt do it.
[18:14:27] <SkullWorks_PGAB> My skype dies in 30 days - and I hate the new pricing structure since ebay bought them out
[18:15:30] <SkullWorks_PGAB> oops too many open windows...
[19:46:30] <Roguish> hello all. quick question for the HAL guys out there.
[19:47:05] <Roguish> what's the procedure for getting a halmeter to run in a seperate terminal window while emc is running?
[19:48:11] <SWPadnos> halmeter doesn't run in a terminal
[19:48:33] <SWPadnos> but, if you're in a separate terminal, you type "halmeter"
[19:48:50] <SWPadnos> or "halmeter &" if you want to get your prompt back
[19:49:10] <Roguish> X_rpm
[19:49:20] <Roguish> wrong paste.
[19:49:29] <SWPadnos> if you're using run-in-place, then you need to source the correct emc-environment in any shell you want to run halmeter that way
[19:49:35] <Roguish> RTAPI: ERROR: version mismatch 533 vs 531
[19:49:37] <Roguish> HAL: ERROR: rtapi init failed
[19:49:48] <Roguish> doh
[19:49:55] <SWPadnos> you need to source the correct emc-environment ;)
[19:50:53] <Roguish> you just reminded me of that. thanks. now, can i put a couple of commands in a script to get serveral meters up at once?
[19:51:05] <SWPadnos> halmeter &
[19:51:07] <SWPadnos> repeat
[19:52:02] <Roguish> can i specify the pin on the halmeter line?
[19:52:04] <SWPadnos> if you want smaller halmeter windows, and you know what you want them to monitor, you can also use halmeter -s {pin|sig|param} <name-of-thing>
[19:52:28] <Roguish> halmeter
[19:52:31] <SWPadnos> the "-s" makes a small window (no signal name or exit buttons))
[19:52:50] <SWPadnos> and sig/pin/param <name> makes it monitor that signal/pin/param
[19:53:58] <Roguish> halmeter -s axis.0.f-error ????
[19:55:02] <Roguish> $ ERROR: 'axis.0.f-error' is not a valid probe type
[19:55:36] <alex_joni> a f-error is not something you want to look at with halmeter anyways
[19:55:41] <alex_joni> you probably want halscope for that
[19:55:52] <alex_joni> but the error suggests you probably have the name wrong
[19:56:29] <Roguish> got it. halmeter -s param axis.0.f-error &
[19:57:29] <Roguish> cool.
[19:59:07] <Gamma-x> anyone know of an anilam crusader upgrade to emc?
[20:00:18] <alex_joni> anilam crusader upgrade?
[20:01:33] <alex_joni> http://www.rittercnc.com/cnc/Re-Controls-Upgrading-Idea-129-.htm
[20:02:12] <Gamma-x> !seen rogern
[20:02:16] <Gamma-x> maby not
[20:02:40] <Gamma-x> pretty much
[20:04:04] <Gamma-x> but there isnt anything on that page about the new system
[20:04:25] <alex_joni> giyf ;)
[20:04:32] <alex_joni> or maybe drop an email to the users list
[20:04:45] <alex_joni> but you probably want to ask a more specific question
[20:05:22] <Gamma-x> well i have a supermax ycm-18 with anilam crusader 2.
[20:05:40] <Gamma-x> I want to ifnd out if anyone has posted how they have upgraded to emc
[20:11:06] <Gamma-x> im on the mailing lsit, how can i send out a new topic?
[20:16:56] <alex_joni> it's a mailing list.. you send an email?
[20:17:19] <alex_joni> I'm sure there are graphical instructions out there that teach you step by step how to send an email
[20:17:40] <DanielFalck> Gamma-x: are you receiving the list mail in digest form?
[20:17:58] <Gamma-x> uhhh they come to me one by one
[20:18:10] <alex_joni> imagine that..
[20:18:11] <Gamma-x> alex_joni, haha thanks
[20:18:16] <alex_joni> they might have been sent one by one :)
[20:19:34] <Gamma-x> hey i never made a post on there haha. gimme a break
[20:20:29] <alex_joni> Gamma-x: sorry.. just pulling your leg ;)
[20:22:17] <Gamma-x> haha its cool
[20:30:36] <alex_joni> good night all :)
[20:31:36] <OoBIGeye> godnight
[20:31:47] <OoBIGeye> *good
[21:01:27] <Gamma-x> how can i search the emc mailing list?
[21:02:43] <fenn> search.gmane.org
[21:03:32] <fenn> put linux.distributions.emc.user in the "group" box
[21:13:45] <Paragon> Hello All. Not sure if you can help but I saw a mill yesterday but can't remember what make it was. I do recall seeing 00 above the spindle. Any ideas?
[21:16:04] <fragalot> lol
[21:16:23] <fragalot> I fear that that is bit too little information..
[21:16:48] <Paragon> lol thought it might be ... :-)
[21:17:21] <Paragon> I am just going through the mills listed on lathes.co.uk in the hope to recognise it.
[21:27:43] <Paragon> I think I've found it ... http://www.lathes.co.uk/elliottmillers/page2.html
[21:44:39] <Gamma-x> would anyone say that emc and mach3 are close to being the same?
[21:45:06] <cradek> maybe if they were clueless or trolling
[21:45:54] <Gamma-x> cradek, hey!
[21:46:19] <Gamma-x> i found a guy online who pretty much did what i am trying to do but he used mach 3,
[21:46:26] <BigJohnT> * BigJohnT smacks Gamma-x
[21:46:44] <Gamma-x> missory!
[21:46:55] <cradek> have you still not been able to tune your servos?
[21:47:09] <Gamma-x> hell no.
[21:47:26] <cradek> remind me what's the servo, amp, and encoder configuration?
[21:47:32] <Gamma-x> uhh
[21:48:00] <Gamma-x> west amp's from the original anilam crusader, linear scales, and servos are uhhh
[21:48:12] <Gamma-x> yca i think
[21:48:18] <Gamma-x> the mill is in ny im now down in dc.
[21:48:22] <cradek> oh only linear scales?
[21:48:24] <Gamma-x> hopefully moving back to ny though
[21:48:28] <cradek> no tachs or encoders on the motors?
[21:48:30] <Gamma-x> yup only linear scales
[21:48:34] <Gamma-x> they have tachs
[21:48:38] <Gamma-x> attached to the amps
[21:48:40] <cradek> ok tachs go to the amps?
[21:48:52] <Gamma-x> yup
[21:49:17] <cradek> do the amps and tachs work right?
[21:49:27] <Gamma-x> http://www.cnczone.com/forums/showthread.php?t=52925
[21:49:32] <Gamma-x> cradek, yes they do
[21:49:54] <Gamma-x> they worked good previously before dismantling the original control.
[21:49:59] <cradek> then it seems like you should be able to tune it
[21:50:33] <Gamma-x> my problem i was having was when i had everything hooked up i would hit a co ordinate and the table would just take off and not stop so i had to hit e stop every time
[21:50:34] <cradek> since it's velocity mode the tuning will be mostly P and FF1
[21:50:57] <cradek> what steps did you take to troubleshoot that? doesn't sound like a tuning problem.
[21:51:21] <Gamma-x> cradek, honestly i dont remember,
[21:51:33] <cradek> hm, hard to help then.
[21:51:37] <Gamma-x> the scales worked properly
[21:51:44] <cradek> sounds like maybe missing or reversed position feedback
[21:51:45] <Gamma-x> cradek, what should I have done then?
[21:51:58] <Gamma-x> position feedback... what delivers that?
[21:52:01] <cradek> systematically troubleshooted
[21:52:05] <cradek> the scales
[21:52:49] <Gamma-x> hmmm they were hooked up to the mesa cards, and as far as I could tell.. properly
[21:52:54] <cradek> emc should have ferrored and turned off the amps right away if it ran away
[21:53:11] <Gamma-x> they gave proper information for locations.
[21:53:24] <Gamma-x> i had a closed loop e stop system
[21:53:26] <cradek> runaway could be caused by the tachs being hooked backward
[21:53:42] <cradek> anything could be wrong - you have to narrow it down
[21:53:42] <Gamma-x> but i didnt touch them when messing with the control.
[21:54:04] <cradek> backward with respect to the commanded velocity, which you did replace
[21:54:15] <Gamma-x> uhhh
[21:54:20] <Gamma-x> in english? lol
[21:54:36] <cradek> you should unhook the velocity command and use a 1.5v battery and see if it moves at a nice even velocity
[21:54:59] <cradek> I suspect you have a very basic problem that has nothing to do with software
[21:55:10] <Gamma-x> muther. fuFger
[21:55:17] <Gamma-x> ur prolly right
[21:55:38] <Gamma-x> do i hook both ends of the tach onto the battery?
[21:55:41] <cradek> verify one part at a time, start at the most basic, then move on up
[21:56:12] <cradek> the tach goes to the amp. that's "actual velocity" input. the amp connection that goes to the mesa card is "commanded velocity"
[21:56:17] <cradek> so the amp has two velocity inputs
[21:56:22] <cradek> it tries to make them MATCH
[21:56:32] <cradek> that's the whole idea of tach feedback
[21:56:37] <cradek> start there -- make sure it works right
[21:56:51] <cradek> then hook to the mesa card, and have the mesa command the 1 volt instead of your battery
[21:56:55] <cradek> see if it still works
[21:57:00] <cradek> if it doesn't, stop, figure out why
[21:57:13] <cradek> if it doesn't work there, it might be grounding for instance.
[21:57:30] <cradek> then build on that
[21:57:45] <cradek> move on to position feedback, make sure emc sees it, make sure it's the right direction
[21:58:11] <Gamma-x> ok sweet
[21:58:15] <Gamma-x> thanks
[21:58:20] <cradek> no problem. just remember to start at the bottom
[21:58:33] <cradek> if any of those things are wrong, you can change emc pid settings for 100 years and it won't work
[21:58:44] <Gamma-x> ahh
[21:58:53] <Gamma-x> im gunna say that ur prolly right
[21:59:04] <Gamma-x> cause pid wasnt doing anything for me
[21:59:48] <Gamma-x> where do i hook that battery to though?
[21:59:59] <cradek> the commanded velocity input of the amp
[22:00:24] <cradek> say full velocity is 200 ipm (guessing). that's +10V. then +1V will be 20 ipm.
[22:00:43] <cradek> if it doesn't move at a nice smooth 20 ipm, something is wrong in the amps/tachs/power supply/motor.
[22:01:17] <cradek> before you do this, get your limit switches and estop right. it's SO EASY to have it run away. make sure they will turn off the amp.
[22:01:41] <cradek> I have to run
[22:01:47] <cradek> I hope you get it
[22:01:57] <Gamma-x> well lucklily like i said the limits are hooked directly to the amps/estop
[22:02:08] <Gamma-x> pull the estop or it hits a limit and the whole thing shuts down
[22:02:13] <cradek> you could even make a plan and take notes. then if you ask for help you can say EXACTLY what you did
[22:02:14] <Gamma-x> than ks cradek
[22:02:20] <cradek> then you can get the best possible answer
[22:02:26] <Gamma-x> cradek, ur right
[22:02:28] <Gamma-x> but im 22....
[22:02:32] <Gamma-x> low attention span
[22:02:35] <cradek> someone might see a step missing from your plan and you can go try that.
[22:02:48] <cradek> can't be that low because it seems like you've been working on this for a year.
[22:02:58] <cradek> age doesn't mean anything
[22:03:05] <fenn> Gamma-x: i bet you dont know how to knit either
[22:03:10] <Gamma-x> true but i like to use it as an excuse lol
[22:03:13] <Gamma-x> hahaha
[22:03:25] <Gamma-x> knitting is not a specialty or myne sorry
[22:03:40] <fenn> http://www.nataliedee.com/090208/you-are-free-to-learn-how-to-do-stuff-at-any-time.jpg
[22:03:41] <cradek> might work somewhere, but not online - fenn is 93 and I'm 11, and we're both fine
[22:03:53] <Gamma-x> haha
[22:04:17] <cradek> fenn: nice
[22:04:19] <cradek> and so true
[22:04:38] <cradek> bbl.