#emc | Logs for 2009-05-17

[00:12:22] <robin_sz> blessed are the paranoid, for they test their gcode
[00:13:19] <archivist> and fewer war wounds in the machines
[00:13:34] <robin_sz> you have no sense of fun
[00:15:11] <anonimasu> yeah praying and pressing "NC START" is always fun
[00:15:47] <robin_sz> whats the worst that can happen?
[00:16:01] <anonimasu> crooked spindle
[00:16:11] <anonimasu> loose parts..
[00:16:14] <archivist> broken bits
[00:16:23] <anonimasu> parts through ou..
[00:16:25] <anonimasu> you..
[00:16:26] <archivist> a screaming boss
[00:16:28] <robin_sz> but think of the youtube fame!!!!
[00:16:38] <anonimasu> um.. actually death..
[00:16:43] <anonimasu> :)
[00:16:57] <robin_sz> so long as its not yor own ...
[00:17:33] <anonimasu> nigth
[00:17:35] <anonimasu> err night
[00:17:42] <robin_sz> yes, that
[00:49:59] <Skullworks> whats the worst that can happen - turret rapids into the tailstock cracking the ball nut mount - $56K repair cost.
[00:55:37] <robin_sz> just make sure you have a webcam set up
[00:56:24] <robin_sz> a service tech told me a good rule of thumb ... $1000 for each 1000 rpm on the spindle when the accident happens
[02:18:54] <tomp> tomp is now known as tomp4
[02:19:16] <tomp4> tomp4 is now known as tomp2
[02:19:41] <tomp2> hmm, i cant be tomp3?? //nick refuses
[02:21:45] <SWPadnos> tomp2, it looks like you're still connected as tomp3 somewhere
[02:22:18] <SWPadnos> if you're sure you have no active connections, you can kick the other instance using the nickserv "ghost" command
[02:22:25] <SWPadnos> other active connections ...
[02:22:28] <tomp2> aha! thx... hmmm did i leave a computer in thailand? in taiwan? ;)
[02:22:45] <SWPadnos> no, it's only been inactive for about 3 days :)
[02:23:05] <SWPadnos> ===tomp3 <n=tomp@> ?purple?
[02:23:07] <SWPadnos> ===tomp3: attached to irc.freenode.net ?http://freenode.net/?
[02:23:08] <SWPadnos> ===tomp3: idle for 3 days, 2 hours, 24 minutes, 21 seconds (on since Wednesday, May 13, 2009 7:55:37 PM)
[02:23:10] <SWPadnos> ---End of WHOIS information for tomp3.
[02:23:43] <tomp2> eh? so its an inactive connection... how do i prevent that ( i use //part usually )
[02:24:04] <SWPadnos> I don't know why it's there. there could be a computer that's still on somewhere
[02:25:18] <tomp2> thx ( mystery to me, tho the hotel and factory aren't 65... )
[02:25:37] <SWPadnos> it's the same as your current IP
[02:30:25] <tomp2> really ?? ifconfig sez im on 179... i dont really understand all this tho
[02:30:45] <SWPadnos> [22:19:39]-->|tomp (n=tomp@ has joined #emc
[02:31:23] <tomp2> yeah, i read that, and ifconfig sez inet addr: so i dunno
[02:31:42] <SWPadnos> uh - do you have NAT anywhere?
[02:32:14] <tomp2> not that i know of, i startup wlan0 then restart networking and I'm in the air
[02:32:15] <SWPadnos> 172.17 is in a class B private network
[02:32:22] <SWPadnos> http://en.wikipedia.org/wiki/Private_network
[02:33:48] <tomp2> ok, that must be the hotel ( where i am ) if this is a problem, is there a way for me to stop it?
[02:34:04] <SWPadnos> your IP address isn't a problem
[02:34:14] <tomp2> the ghost i mean
[02:34:32] <SWPadnos> oh, /msg nickserv ghose tomp3 <yourpassword>
[02:34:35] <SWPadnos> err
[02:34:39] <SWPadnos> s/ghose/ghost/
[02:36:48] <tomp2> no password. i remember years ago on this irc, i register, i set pswd, i can never log in again. several tries. now i just dont register. and here i is.
[02:37:15] <SWPadnos> oh. in that case I don't know how to get rid of the tomp3
[02:37:43] <tomp2> thx hope its no prob
[02:37:46] <SWPadnos> you'd think it would have timed out or had its connection reset or something
[02:37:52] <SWPadnos> shouldn't be, but who knows
[02:38:15] <SWPadnos> I suppose I should try to force myself to sleep. I have to get up at 4:30 tomorrow morning
[02:38:22] <SWPadnos> I sense a topic change coming
[02:38:22] <jepler> jepler has changed the topic to: Welcome! EMC (Enhanced Machine Controller) is a linux-based opensource CNC control. | Latest release: EMC 2.3.\1 | http://www.linuxcnc.org | http://wiki.linuxcnc.org | Channel logged by logger_emc
[02:38:34] <jepler> jepler has changed the topic to: Welcome! EMC (Enhanced Machine Controller) is a linux-based opensource CNC control. | Latest release: EMC 2.3.1 | http://www.linuxcnc.org | http://wiki.linuxcnc.org | Channel logged by logger_emc
[02:38:41] <tomp2> clair-o-voyant
[02:38:43] <SWPadnos> cool!
[02:43:04] <tomp2> this is cool ( i was looking for cad for wiring ) http://www.pdfee.com/ has free pdf wiring diagrams for many cars, motorcycles, boats...
[02:43:40] <roh> oh.. emc 2.3.1 updates....
[02:44:36] <roh> hm.. nothing fancy for me
[02:45:11] <SWPadnos> tomp2, if you find a good cabinet wiring program, let me know
[02:45:47] <SWPadnos> so far everyone I know uses whatever CAD package they're used to (AutoCAD, some PCB thing, Curel Draw ...)
[02:45:54] <SWPadnos> Corel
[02:46:51] <tomp2> i got autocad electrical, its just huge ( and w$ )
[02:47:03] <SWPadnos> and $$
[02:48:27] <tomp2> i came across interkonnecto, but gave up ( its in esperanto , oooh a brand new dead language )
[02:48:33] <roh> is there something like a g-code-execution-ETA plugin already?
[02:49:01] <SWPadnos> file->properties will give you a run-time estimate
[02:49:05] <SWPadnos> in AXIS
[02:49:29] <mozmck> kicad is a pretty good schematic/pcb layout CAD
[02:49:49] <mozmck> would probably work well for wiring diagrams I guess.
[02:50:00] <SWPadnos> maybe, but probably not
[02:50:16] <SWPadnos> I have Altium Designer, and before that had Protel Advanced Schematic
[02:50:16] <roh> SWPadnos ah.. nice.. that already helps a bit. but something that counts down would be nice also. since it depends on feed override etc
[02:50:25] <SWPadnos> they're not quite the right thing
[02:50:41] <SWPadnos> if you use FO, then estimates are useless anyway ;)
[02:50:44] <roh> * roh ended up with using gnumeric for cabling plans
[02:50:53] <tomp2> mozmck, thx, wiring diagrams are a bit different, its half a cartoon and half a database
[02:51:37] <mozmck> oh, ok. don't have any experience with them.
[02:51:37] <SWPadnos> and things like wire numbers, cut lists, termination types, etc. are the real problem
[02:51:44] <roh> stuff like it-conferencing-setups with a few thousand rj45 connections, pri and bri isdn, pots. ethernet in 100 and 1000mbit with and without poe.. etc
[02:51:44] <tomp2> i like in orcad, the ability to click on a signal, and all the wire lights up , even inside connected prints
[02:52:11] <SWPadnos> or just the wires in a CNC control cabinet, for example ;)
[02:52:24] <mozmck> tomp2: kicad does that
[02:52:28] <tomp2> right, thats my app now
[02:52:35] <tomp2> mozmck, will look thx
[02:53:21] <mozmck> http://kicad.sourceforge.net/wiki/index.php/Main_Page
[02:53:24] <SWPadnos> I wonder what happens if 10 processes all try to mkdir -p the same directory at the same time
[02:53:42] <roh> SWPadnos i think there is no universal solution without figuring out a naming and classification-scheme for every usecase
[02:53:53] <tomp2> (does anyone sell covers for these damn synaptic pads where my thumbs keep sending the cursor shooting off... gimme that bad cd! i'm gonna glue it over that stupid thing )
[02:54:02] <SWPadnos> roh, you mean for cabinet wiring?
[02:54:20] <SWPadnos> tomp2, turn off multi-touch if it's on
[02:54:25] <SWPadnos> I'm not sure how to do that
[02:54:27] <roh> SWPadnos aswell. also for smaller stuff which is beyond the scope of a schematic
[02:55:04] <roh> anything thats used for nonstatic setups e.g. patchpanels of some sort. could be also video or sensoring stuff
[02:55:11] <SWPadnos> my laptop pad can register 3 touches, but it reports the "main" one as the center of the two or three recognized spots
[02:55:16] <tomp2> today i just decided .... this area is 4xx, this area is 3xx, execpt for those few wires already numbered, we're inventing new ones
[02:55:31] <roh> usually one ends up with tables in some worksheet
[02:55:32] <tomp2> re: touchpad, thx, will try
[02:55:42] <SWPadnos> so if you accidentally touch the bottom of the pad, it's like you just accelerated very fast over half the area :)
[02:56:14] <SWPadnos> roh, ok. I'm thinking of stuff more like CNC wiring (the reason I originally searched)
[02:56:41] <SWPadnos> I also had a project where I designed a board and wanted to make a diagram showing how it had to connect inside the machine it was made for
[02:56:56] <roh> eventually using a schematic-sw and using some kind of 'block schematic style' would fit
[02:56:56] <tomp2> way cool, i can disable the sucker, i got a mouse anyway thank you thank you thank you
[02:57:02] <SWPadnos> I ended up annotating an assembly print with labels
[02:57:07] <SWPadnos> heh
[02:58:16] <SWPadnos> that has pitfalls. for example, the connectors on the PC board are components, and I don't know that they can be marked as interconnects
[02:58:57] <SWPadnos> you could make a symbol for a completed PCB which has all the ports on it, but that isn't driven by the actual design, so it's prone to errors
[02:59:30] <SWPadnos> if you try to make a sheet symbol with a compete board and ports, then a parts list will (generally) include all of the components that are on the board
[02:59:52] <SWPadnos> you can use a flat wiring diagram, but that doesn't work so well for large machines
[03:00:00] <tomp2> i've not had lots of experience with the phoenix type din rail stuff for wiring. they multi level connectors with shorting bars etc, 6mm wide relay sockets with plugin ssr/mechanical relays, fuse blocks etc. but htere ought be a lego-like cad for that
[03:01:11] <SWPadnos> (we did a power supply that had about 30 electronic modules in 4 compartments of a large enclosure - each panel would have been great for a sheet or two, then go up a level to see how the panels connect, then up a level to see interconnects between compartments, then up another level to see how to connect the machine to the rest of the world)
[03:01:34] <tomp2> sounds like orcad ( up level )
[03:01:40] <SWPadnos> does Phoenix have anything for download?
[03:01:47] <tomp2> didnt see any
[03:01:54] <SWPadnos> or Wago, AB, Weidmuller ...
[03:02:06] <tomp2> wego might,
[03:02:13] <tomp2> thx, will look
[03:02:18] <SWPadnos> sure, Orcad can do hierarchies, as can Altium or other high end PCB CAD
[03:02:42] <SWPadnos> but the component lists don't usually include the wires in a schematic
[03:02:48] <SWPadnos> I guess that's the main difference
[03:03:15] <leito> hello everybody, i have problems with the comp utility, i need some packages for make it work but i can't find where they are.. i tried looking into the mailing list but nothing i will be thanked if anyone knows how can i make comp work ... thanks in advance :)
[03:03:20] <SWPadnos> a cabinet wiring diagram is concerned with the interconnections, a schematic doesn't treat those as components, since they're expected to go on a PC board
[03:03:45] <SWPadnos> leito, http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2#Preparing_Ubuntu_to_compile_emc2
[03:03:54] <SWPadnos> actually, don't use that one
[03:04:30] <leito> thank you SWPadnos i will check that :)
[03:04:47] <SWPadnos> wait a sec, there's a better (and easier) set of instructions :)
[03:05:04] <SWPadnos> I think you only need to "apt-get build-dep emc2"
[03:05:08] <SWPadnos> err, sudo apt-get ...
[03:05:13] <SWPadnos> assuming you're on Ubuntu
[03:05:29] <SWPadnos> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?ContributedComponents#How_to_compile_and_install_a_component
[03:06:07] <SWPadnos> ok, time for bed. see you all later
[03:06:08] <leito> yes i'm on ubuntu, i donwloaded the live cd iso, but is updated with the 2.3 beta i have to update to the 2.3 final version
[03:06:18] <leito> thank you agian goodby :)
[03:06:21] <leito> bye
[03:06:52] <tomp2> gnite
[03:07:07] <tomp2> thx
[03:13:23] <tomp2> wago proserve looks promising for cabinet wiring (free, close src, windows based, and wago products only) i begin with the manual, i can read the pdf on linux.
[03:14:13] <tomp2> no idea how itd connect to a drive or p/s ( not wago ) but still reading
[03:20:49] <tomp2> its got user devices, imports from excel, looks good enuf to try ( 650meg dload ! )
[03:23:24] <mozmck> I don't understand center format arcs in Gcode
[03:26:06] <tomp2> you are someplace. you want to move in an arc. you describe where the destination of the end of the arc, and where the center of the arc is. The dest is either absolute or 'from here'. the center is always . from here'
[03:26:18] <tomp2> the end is XY, the center is IJ
[03:26:28] <tomp2> ( in the xy plane )
[03:26:45] <tomp2> the ij is where you swing from ;)
[03:32:52] <mozmck> I think I get it...
[03:34:19] <mozmck> I see, the start point is the current location. That's what was throwing me off. I was trying to figure out where the other end of the arc was
[03:34:48] <tomp2> yes the arc is from where you are, always
[03:34:49] <L84Supper> SWPandos : hi, did you ever get the video working well enough for that card?
[03:35:10] <tomp2> i think hes asleep (or packing)
[03:36:31] <tomp2> mozmck: and you still need to say where the end is. one way is like latitude and longitude, the other way is 'from here' ( absolute/G90 and relative/G91 resp. )
[03:36:49] <roh> a friend just threw 'stardraw.com' into the round (for wiring stuff)
[03:37:30] <mozmck> yeah, so one endpoint is 'here', the other is given and the center is given.
[03:37:42] <mozmck> absolute is more common isn't it?
[03:39:16] <tomp2> i duno if G90 is more common, but I use it a lot, and use G91 in subroutine ( so they can be be from 'wherever' )
[03:40:52] <tomp2> i think of G90 as 'locations' and G91 as 'directions' , G90 sez goto chicago, G91 sez go east 200 miles
[03:43:00] <tomp2> dang, kicad is only 70meg and i been dloading it for an hour! smeggy connection here.
[03:44:31] <mozmck> 70 meg? I didn't remember is being anywhere near that big.
[03:45:07] <tomp2> 70.2
[03:45:29] <mozmck> for windows?
[03:45:47] <Skullworks> Ha - my sticker came today so my Mustang GT will be legal again - Now I can make a speed run to the fest... maybe.
[03:45:50] <tomp2> i dont think so, its a tgz
[03:46:36] <mozmck> hmm. It has a lot of libraries and stuff with it I think. I compile my own.
[03:47:18] <tomp2> be careful on those long stretches in nebraska, tempting to open that baby up.
[03:48:08] <mozmck> I don't have any experience with any other EDA programs, but i've seen others compare it very favorably to many of the commercial programs.
[03:49:07] <tomp2> while in lincoln, stop in uni and check out the manufacturing labs. they do automation, edm, cells, interesting stuff.
[03:49:29] <tomp2> prof rajakur ?
[03:59:02] <tomp2> wow i cant find any of the old links to the studies done there. but, i can find loads of patent references! from the guys who worked with Rajakur on those projects. i thought that stuff belonged to the uni or the state.
[03:59:34] <tomp2> oh well, nitey nite
[10:32:41] <JanVanGilsen> Poincare: I got the PUMA moving, at least one joint.. :-)
[11:57:09] <alex_joni> JanVanGilsen: cool
[12:03:18] <JanVanGilsen> Still need to figure out the homing, the puma has multi-turn potentiometers on each joint to estimate it's absolute position, this estimation is used to calculate the position of the next index pulse
[12:04:02] <BigJohnT> hmm, I get the hostmot2 update but not 2.3.1 when I reload the package information on 8.04 guess it takes a while
[12:08:01] <JanVanGilsen> That would require a hack to get the [AXIS]HOME parameters as pins in hal ...
[12:10:52] <JanVanGilsen> I could use a window on the pot signal to simulate a home switch, but that can result in large unpredictable moves of the arm
[12:13:59] <tomp> tomp is now known as tomp4
[12:20:48] <alex_joni> BigJohnT: I think you are correct
[12:21:19] <alex_joni> http://linuxcnc.org/hardy/dists/hardy/emc2.3/binary-i386/Packages doesn't contain info for emc2 or emc2-dev 2.3.1
[12:21:28] <alex_joni> jepler: ^^
[13:00:53] <jepler> alex_joni: weird, I'll look into it
[13:06:20] <jepler> alex_joni: the cause was fumble-fingers on my end .. should be fixed very shortly.
[13:23:03] <Poincare> JanVanGilsen: what's the PUMA ???
[13:25:53] <jet> jet is now known as bjt-plasma
[13:26:21] <bjt-plasma> I get "W: Failed to fetch http://www.linuxcnc.org/hardy/dists/hardy/emc2.3/binary-i386/emc2_2.3.1_i386.deb
[13:26:23] <bjt-plasma> 404 Not Found" when I did the update
[13:26:24] <JanVanGilsen> my robot :)
[13:26:57] <JanVanGilsen> I did mention it a couple of moths ago
[13:29:01] <JanVanGilsen> *months
[13:40:55] <jepler> bjt-plasma: give it a try again please
[13:41:12] <jepler> I think when you tried it I was only half-done fixing what I'd fouled up last night
[13:45:01] <bjt-plasma> It is downloading now jepler
[13:45:28] <bjt-plasma> I usually foul up first thing in the morning just to get that out of the way :)
[13:45:53] <jepler> good
[13:45:57] <archivist> does 2.3 have 5 axis jog
[13:52:44] <JanVanGilsen> hm2-servo example config: the axis.n.index-enable should be linked to hm2_[HOSTMOT2](BOARD).0.encoder.nn.index-enable
[13:53:55] <JanVanGilsen> It did take me some time to figure out why the homing didn't work :)
[16:15:42] <bjt-plasma> I just got done copying the info from my working plasma ini to my 5i20 ini and to my surprise it works sorta :/
[16:16:22] <bjt-plasma> if I jog at 396 IPM everything is fine if I jog at max velocity of 405 IPM I get a Joint n following error
[16:17:34] <bjt-plasma> If I do a G0 Xn it runs along to the end with no error
[16:17:54] <bjt-plasma> and I don't get the error until I release the jog button
[16:18:35] <bjt-plasma> doesn't matter if I use the jog button or the keyboard arrow keys
[16:23:01] <bjt-plasma> If I raise the axis max velocity above the traj max velocity it does not error
[16:29:09] <bjt-plasma> is this the expected behavior?
[16:29:56] <BigJohnT> * BigJohnT wanders back outside to bask in the sun :)
[16:51:03] <bjt-plasma> seems like if the axis max velocity is set above 1/2 of the traj max velocity and you jog at max velocity you get "joint n following error"
[16:51:32] <bjt-plasma> above 1/2 but below traj max velocity
[16:51:43] <bjt-plasma> or equal to traj max velocity
[16:53:10] <Optic> so, is there a good, reasonably priced CAM thing that will reliably make good gcode from DXF?
[16:55:07] <bjt-plasma> and it does the same in tkemc...
[16:56:26] <bjt-plasma> good, cheap, and reliable never seem to fit in the same sentence :)
[16:57:24] <archivist> what.... even EMC2 :)
[16:58:59] <Optic> i'm just looking for something that can do nice simple toolpaths
[16:59:14] <Optic> i've been using dxf2gcode python and it's pretty good. but sometimes i makes broken g2's
[16:59:21] <archivist> hand hack the gcode
[16:59:29] <Optic> i have been :)
[16:59:39] <Optic> i think it makes rounding errors
[16:59:51] <Optic> but qcad -> dxf2gcode works ok
[16:59:59] <archivist> floats are a common problem
[17:00:04] <Optic> i've been quite impressed with qcad
[17:01:27] <cradek> bjt-plasma: sounds like stepgen headroom problem
[17:01:28] <Optic> maybe i should hack on dxf2gcode and clean it up a bit
[17:01:29] <bjt-plasma> EMC is free
[17:02:21] <bjt-plasma> cradek: would that only affect the jog and not a rapid?
[17:02:37] <cradek> jog uses higher acceleration
[17:02:49] <cradek> if you program g61 you might see the same problem with g0
[17:03:35] <Optic> archivist: i should port dxf2gcode to python 2.6 actually, I think it does all fixed-point arbitrary precision math by default
[17:04:24] <bjt-plasma> cradek: yep it is there with g0 with g61
[17:05:04] <cradek> yep it's just stepgen headroom I bet
[17:05:37] <bjt-plasma> so my max vel in traj needs to be over twice any axis max vel?
[17:05:46] <cradek> no
[17:05:50] <archivist> Optic, make sure touching endpoints actually have the same value, Ive seen insane small differences at the connections
[17:06:05] <cradek> easiest thing to do is not put any vel,acc in traj
[17:06:17] <Optic> archivist: I guess I should make sure that qcad is making a sane dxf first
[17:06:23] <cradek> those are for mostly useful for nontrivkins machines
[17:06:28] <bjt-plasma> ok
[17:07:22] <archivist> Optic, we had to fix an old dos package years ago that had bad dxf out, I may still have the script
[17:08:23] <Optic> i think qcad is doing ok
[17:09:18] <archivist> our supplier for laser cut steel was not till we fixed the dxf's
[17:10:48] <bjt-plasma> cradek: I still get the error if I put max vel in display section
[17:11:14] <bjt-plasma> otherwise the slider only goes as far as the slowest axis
[17:12:11] <cradek> I don't understand
[17:13:05] <bjt-plasma> if I just remove any vel acc in traj then the jog speed slider is maxed at 60 the speed of the slowest axis
[17:13:37] <cradek> but what error?
[17:14:09] <bjt-plasma> if I add MAX_LINEAR_VELOCITY = 7 to the DISPLAY section so the slider will go to 420 when I jog at 420 I get the joint following error for that axis
[17:14:45] <cradek> ok, what is your stepgen headroom on that axis?
[17:15:54] <bjt-plasma> I don't see that for that axis
[17:16:19] <cradek> well you need some
[17:16:26] <cradek> pastebin hal and ini?
[17:16:35] <bjt-plasma> one moment
[17:17:19] <bjt-plasma> btw, these are based on the 5i20 stepper config
[17:18:22] <bjt-plasma> http://pastebin.ca/1425601
[17:18:51] <bjt-plasma> and http://pastebin.ca/1425603
[17:18:59] <cradek> did you get them from 2.3.1? I see stepgen headroom for hm2-stepper was added in 2.3.1
[17:19:02] <cradek> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/configs/hm2-stepper/5i20.ini.diff?r1=1.1;r2=;only_with_tag=RELEASE_2_3_1
[17:19:19] <bjt-plasma> I thought I did...
[17:19:58] <bjt-plasma> but I guess I did not :(
[17:20:05] <cradek> setp hm2_[HOSTMOT2](BOARD).0.stepgen.00.maxaccel [AXIS_0]MAX_ACCELERATION
[17:20:08] <cradek> ^ this is wrong
[17:20:32] <cradek> if those are the same, it will just barely not work, like you are seeing
[17:21:22] <cradek> here is the relevant hal change in 2.3.1 too: http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/configs/hm2-stepper/hm2-stepper.hal.diff?r1=1.1;r2=;only_with_tag=RELEASE_2_3_1
[17:22:15] <bjt-plasma> I'll start over with a for sure 2.3.1 config
[17:22:43] <bjt-plasma> cradek: thanks
[17:22:44] <cradek> bleeding edge :-)
[17:22:52] <cradek> welcome
[17:23:03] <cradek> bbl
[17:23:15] <bjt-plasma> soon as I get it worked out I'm gonna make some sparks
[18:30:26] <bjt-plasma> cradek: thanks it works great now :)
[18:52:28] <cradek> yay
[19:15:08] <bglackin> Optic: I loaded HeeksCAD and HeeksCNC last eve. It looks pretty good so far and has distros for Ubuntu and Windows
[19:20:00] <bjt-plasma_> bjt-plasma_ is now known as bjt-plasma
[19:34:49] <BigJohnT_> BigJohnT_ is now known as BigJohnT
[19:50:57] <L84Supper> when trying to build emc for unbuntu 9.04 it breaks with : rtai_shm wasn't found or something
[19:51:34] <jmkasunich> did you build a realtime (RTAI) patched kernel for 9.04?
[19:51:40] <L84Supper> yes
[19:51:50] <L84Supper> why it was odd
[19:52:01] <L84Supper> will look again later today
[19:52:54] <L84Supper> probably new ubuntu strangeness
[19:53:29] <L84Supper> I'll work it out and post a package
[19:54:48] <L84Supper> is there any breakage of EMC on the latest kernels?
[19:55:29] <jmkasunich> dunno, I'm still using Ubuntu 6.06
[19:55:42] <L84Supper> safe and sound :)
[19:56:01] <jmkasunich> I've never been accused of being an early adopter
[19:57:14] <L84Supper> I'm sure your 386 has plenty of good years left :)
[19:57:41] <jmkasunich> ;-)
[19:57:48] <jmkasunich> my hardware is newer than the OS
[19:58:11] <Optic> have any of you used CAM Expert?
[19:58:12] <jmkasunich> I figure, why install the OS more than once on any given computer
[19:58:19] <jmkasunich> Optic: sorry, nope
[19:58:45] <L84Supper> actually trying to compare performance of EMC on the VIA C7 cx700 vs AMD 690/600 with Ubuntu 9.04
[19:59:05] <archivist> Optic, there is also #cam
[19:59:13] <Optic> oh really? :)
[19:59:13] <Optic> hehe
[19:59:37] <Optic> now i'm there!
[19:59:39] <archivist> yeah really
[20:21:16] <tomp> tomp is now known as tomp4
[20:24:06] <bjt-plasma> I'm hijacking the Z position command on my thc and adding in any correction as needed.
[20:24:47] <bjt-plasma> when the torch is turned off and the Z moves up I need to remove any correction but make sure I don't move the torch in the - direction
[20:25:26] <bjt-plasma> can any of you math wizards give me a clue as to where to start?
[20:26:18] <bjt-plasma> I want to remove the correction at a constant amount until the correction == 0
[20:26:52] <bjt-plasma> relative to the speed of the Z
[20:30:24] <bjt-plasma> * bjt-plasma wonders if a Corona will help me figure it out or if Corona is the reason I can't see the answer yet in my head
[20:46:11] <skunkworks> * skunkworks is having leinenkugel leftovers from the reggea fest.
[20:46:52] <skunkworks> * skunkworks should really finish the ridge cap
[20:47:16] <bjt-plasma> are you up to the ridge cap?
[20:53:09] <skunkworks> yes
[20:53:31] <bjt-plasma> sweet
[20:53:35] <skunkworks> yes :)
[21:29:40] <alex_joni> Optic: you need to fiddle with udev some for rtai_shm to work
[21:30:49] <alex_joni> Optic: check http://linuxcnc.org/hardy/dists/hardy/base/source/rtai_3.6.1-linuxcnc.4.tar.gz
[21:31:02] <alex_joni> or the deb, to see how it's done
[21:37:31] <roh> hmm.. is ir normal to have a insane high load on emc2 machines?
[21:37:44] <roh> something in the area of 3-6
[21:38:06] <roh> i guess it has something to do with the rtai tasks
[21:38:09] <alex_joni> roh: that's a bit too high
[21:38:21] <alex_joni> rtai stuff usually doesn't show up in top & co.
[21:38:36] <alex_joni> it's running besides linux, so linux tools to measure load don't catch it
[21:39:22] <roh> alex_joni well.. still it eats cycles on the cpu, which then the sytem cannot use to execute and need to 'delay' execution, thus 'pushing up the visible load' atleast thats what i assumed
[21:39:49] <alex_joni> right, the available time for processes shrinks
[21:40:01] <alex_joni> but you shouldn't have a load over 1 for regular emc2 work
[21:40:03] <roh> load doesnt mean its not idle in linux.. it just means it cannot execute something when it wanted immediately
[21:40:07] <alex_joni> what processor do you have?
[21:40:22] <roh> its an old duron1400 or 1600 or so
[21:40:39] <roh> socket A with a gig ddr or so
[21:40:48] <alex_joni> I run an Athlon XP 1400+, and it's ok
[21:40:48] <roh> works fine, still
[21:40:51] <alex_joni> but on 6.06
[21:41:00] <roh> 8.04.1 with emc 2.3.0 atm
[21:41:19] <alex_joni> 8.04 might be a tad bit much for the duron ;)
[21:41:28] <alex_joni> maybe you can try switchin gto Xubuntu
[21:41:39] <roh> nah.. the os itself is fast and responsive.. no worries there
[21:42:14] <roh> i know my linuxes in detaill.. thats not the problem.. i just havent had much rtai-experience since franky... nobody uses it anymore
[21:42:32] <roh> for 99.99% of the apps -rt is enough
[21:42:42] <roh> (the ingo mollnar stuff in mainline kernel)
[21:43:04] <alex_joni> I'm pseudo-familiar with it ;)
[21:43:30] <alex_joni> it should be ok enough for a servo machine maybe, with HW for interfacing
[21:43:42] <alex_joni> even a mesa stepgen "might" work on -rt some day
[21:44:06] <roh> mesa should work even without if done right
[21:44:21] <alex_joni> I don't think it can
[21:44:26] <roh> since all pwm and timing should/could be done on the fpga fw
[21:44:28] <alex_joni> you still need a servo-thread
[21:44:34] <alex_joni> to update the next position target
[21:44:45] <alex_joni> so the fpga has something to do for the next interval
[21:44:59] <roh> eeeh.. why not do timing in there?
[21:45:12] <roh> and use a cached command pipeline
[21:45:12] <alex_joni> you mean put a processor in the fpga ?
[21:45:19] <alex_joni> you can only cache so much
[21:45:31] <alex_joni> if you want to do threading, then you need realtime feedback
[21:45:37] <roh> no.. just the timing and some simple fifo with bufferlogic
[21:45:39] <alex_joni> same goes for speed-override buttons
[21:45:44] <roh> should be standard macrocells
[21:46:10] <alex_joni> if the user changes speed override, then everything you've cached goes out the window
[21:46:27] <roh> emc would be degraded to 'just view and controll dataflow' instead of timing all and everything in detail.. kinda 'outsourcing'
[21:46:50] <roh> alex_joni if the user changes it, just tell the fpga the updated data.. where is the problem?
[21:46:52] <alex_joni> roh: that means you take about 50% of emc, and move it from SW into HW
[21:47:07] <alex_joni> roh: yeah, but the user expects it to react now
[21:47:11] <roh> its a fpga.. -> still sw. just the execution-domain and arch changes
[21:47:27] <alex_joni> not in 2 seconds when the buffer emptied, or a new one has been generated
[21:47:39] <roh> alex_joni the user has a usb-keyboard so rtai isnt doing anything for <pollingspeed of usb> anyways
[21:47:57] <alex_joni> yeah, but he can have a rt connected jogwheel for controlling speed override
[21:48:11] <roh> even with ps2, all rtai threads are higher prioritised, so any user in or output happens non-rtai anyways.
[21:48:13] <alex_joni> same is more important for threading
[21:48:31] <roh> still msecs are _much_ enough.. -rt does ysec stuff already without killing the rest of the system
[21:48:36] <alex_joni> where you need to sync movement to the signals from the encoder
[21:48:46] <alex_joni> yeah, most of the time
[21:48:52] <alex_joni> but imo that is not good enough
[21:49:24] <roh> ah.. 50-60microsecs worst execution time.. should be enough when you dont need to 'wiggle gpio/printerports' through 3-4 layers anymore
[21:49:32] <alex_joni> and re killing of the system, I've only seen that when the base thread is set too low
[21:49:45] <alex_joni> usually I don't notice any difference between rtai running/not running
[21:49:50] <roh> my base-thread was something like 18000 i think
[21:49:56] <alex_joni> that's pushing it
[21:50:00] <alex_joni> for that machine
[21:50:41] <alex_joni> I bet it'll be a _lot_ different at 20-22 usec
[21:51:05] <alex_joni> but it depends what you need ;)
[21:51:11] <alex_joni> maybe you do need the 18000
[21:51:40] <roh> it was 16000 but i was seeing rtaiÃ-problem-errors which indicate a 'too low' seldomly with it
[21:51:57] <roh> e.g. when rapidly increasing feed override to 120
[21:52:00] <roh> while milling
[21:59:16] <renesis> hey guys
[21:59:35] <renesis> does emc have m-codes for program jumps or what?
[22:00:57] <renesis> like how do i make something loop forever
[22:01:38] <renesis> (i think its m98/m99 on haas thingers, one being a call and the other being a return)
[22:03:00] <renesis> hmm
[22:05:01] <alex_joni> renesis: simple o-word loop
[22:05:15] <alex_joni> just set an impossible condition for ending the loop
[22:05:23] <alex_joni> o100 do
[22:05:36] <alex_joni> o100 while [1 GT 2]
[22:05:40] <alex_joni> or something like that
[22:06:22] <renesis> i cant just do while [1]
[22:07:02] <renesis> last time i did conditional gcode on emc i just remember getting really pissed off =\
[22:34:27] <colin__> why do people even bother with steppers ?
[22:35:18] <colin__> have some on my mini lathe and after 5 mins of using them i hate them already
[23:01:35] <renesis> heh
[23:01:49] <renesis> i did do stuff while [1]
[23:01:56] <renesis> and axis loaded it and now its stuck
[23:01:57] <renesis> =(
[23:05:11] <renesis> how do i get it to load without processing it first
[23:05:30] <SWPLinux> escape?
[23:25:27] <renesis> that worked but now its super laggy in response to keyboard input
[23:25:37] <renesis> like i tap a jog key and it goes half an inch
[23:51:08] <SWPLinux> you should correct the problem in the gcode, and/or put (axis, hide) / (axis, show) around your infinite loop
[23:51:33] <SWPLinux> the hide/show will prevent axis from putting the motion in the preview display
[23:52:00] <SWPLinux> (this will of course make it non-visible until the path has been run once)