#emc | Logs for 2005-10-23

[13:25:59] <Imperator_> alex_joni: Hi Alex
[13:26:13] <alex_joni> hey Martin
[13:26:24] <alex_joni> * alex_joni is working on another tkemc bug :D
[13:26:30] <Imperator_> thats realy strange with my servo card, the Dac is working, but the counters not
[13:26:45] <alex_joni> hmm.. you sure?
[13:27:16] <Imperator_> jep i can mesure the voltage, and it changes, when i chance the value of the HAL pin
[13:27:29] <alex_joni> are you getting only 0's ?
[13:27:30] <alex_joni> on the counters?
[13:27:45] <Imperator_> jep, it doesen't count
[13:27:55] <Imperator_> the scaling is not the problem
[13:28:03] <alex_joni> * alex_joni thinks the counters are not initialized
[13:28:04] <Imperator_> because default scale is 0
[13:28:09] <alex_joni> what counters are on the card?
[13:28:13] <alex_joni> don't look at scale..
[13:28:18] <alex_joni> look at raw counts
[13:28:25] <alex_joni> scale should be 1 (default)
[13:28:36] <alex_joni> because position=raw * scale
[13:28:36] <Imperator_> something special, i don't think there is something to initialise
[13:28:45] <Imperator_> jep
[13:28:49] <alex_joni> if scale is 0 then position is always 0
[13:28:57] <alex_joni> so raw stays at 0 too ?
[13:29:08] <Imperator_> now i have to check my servos, i hope i have not killed the encoders
[13:29:30] <Imperator_> jep it stays, but i have to look exactly on raw counts
[13:30:17] <alex_joni> code looks ok
[13:30:43] <Imperator_> jep
[13:31:05] <Imperator_> it works the last time
[13:31:43] <Imperator_> what is that new parameter update.time ???
[13:32:00] <alex_joni> where is that?
[13:32:02] <Imperator_> is that now exportet for every module ???
[13:32:22] <Imperator_> it is a parameter pin, and i don't have exportet that parameter
[13:33:02] <Imperator_> evoreg.1.update.time and evoreg.1.update.tmax
[13:35:00] <Imperator_> maybe it was ther all the time, but i never noticed it
[13:37:26] <alex_joni> it was
[13:37:35] <alex_joni> it's the time for your exported function
[13:37:37] <alex_joni> hal-internal
[13:37:46] <alex_joni> you can do debugging how fast your code runs
[13:37:54] <alex_joni> and maybe put it in a faster thread if you need
[13:41:05] <alex_joni> ok.. brb
[13:41:08] <alex_joni> booting an BDI
[13:47:19] <alex_joni> back
[13:48:01] <alex_joni> * alex_joni is away: back-later
[14:41:13] <alex_joni> * alex_joni is back
[15:01:17] <alex_joni> anyone who knows tcl around?
[15:01:37] <anonimasu> :)
[15:06:16] <rayh-away> I know just a bit about tickle.
[15:07:29] <rayh-away> rayh-away is now known as rayh
[15:25:55] <Jacky^> Jacky^ is now known as Jacky^afk
[16:05:05] <jmkasunich> * jmkasunich is packing up and heading home... a very good week!
[16:09:02] <Jacky^afk> Jacky^afk is now known as Jacky^
[16:18:23] <CIA-6> 03jmkasunich * 10emc2/configs/mazak_rf.clp: re-entered the ladder program that was lost
[16:18:27] <CIA-6> 03jmkasunich * 10emc2/nc_files/ (3D_Chips.ngc cds.ngc): re-entered the ladder program that was lost
[16:29:04] <steves_logging> steves_logging is now known as steves_stallings
[16:47:44] <fenn> hi dave
[16:47:56] <fenn> did you hear - the mazak is running
[16:48:45] <dan_falck> is it doing tool changes, moving the table, turning on the spindle?
[16:49:15] <fenn> no tool changes, but spindle and motion work
[16:49:32] <fenn> gear change works
[16:49:37] <dan_falck> good
[16:49:48] <rayh> Hi guys. I'll be hanging around but only once in a while.
[16:49:57] <dan_falck> hi ray
[16:49:57] <steves_stallings> and spindle speed and gear change with classic ladder
[16:50:08] <fenn> rayh: you said hydraulics work, does that mean you got tool changes to work?
[16:50:52] <rayh> We have most of the tool change actuators going. Still a bit of ringing to figure out a couple solenoids and prox switches.
[16:51:13] <rayh> Then the work on ladder to make certain that the proper sequences are enforced.
[16:51:21] <rayh> Hi Dan.
[16:51:36] <rayh> Get some guitar playing time on friday?
[16:51:46] <dan_falck> yes. lots
[16:52:13] <rayh> We do have spindle speed in hal. Have to add a bit to ioconfig for emc spindle speed.
[16:52:22] <rayh> We do have gear change as a single hal command.
[16:52:50] <rayh> Will write two little bash scripts to connect this to m101 and m102
[16:53:14] <cncuser> hi rayh, got your partitions fixed yet ? :)
[16:53:21] <rayh> We also have spindle orient but still need some tuning to minimize deadband and
[16:53:35] <rayh> then tool change
[16:53:49] <rayh> I did. cncuser. Thanks for the heads up.
[16:54:07] <cncuser> rayh: still problems with unclean unmounting ?
[16:54:25] <rayh> I've not had much time to play with it but puppy-emc is going to be an awesome addition to the EMC systems.
[16:54:42] <cncuser> rayh: yeah, ok
[16:54:44] <cncuser> gotta go
[16:56:11] <dan_falck> what is puppy-emc?
[16:56:37] <steves_stallings> a smaller cuddly version of Linux running EMC
[16:56:51] <dan_falck> EMC2?
[16:56:56] <steves_stallings> yes
[16:57:05] <dan_falck> cool\
[16:57:18] <steves_stallings> think it has full dev tools too
[16:57:31] <fenn> but no web browser?
[16:58:02] <fenn> or maybe i just dont know what it's called
[16:58:29] <steves_stallings> not sure what baggage got tossed, but it is less than 50 meg ISO image
[16:59:59] <steves_stallings> cncuser (I think) did the work to install rtai and make images
[17:00:07] <fenn> yeah
[17:01:06] <fenn> dillo is a good browser i just discovered and it's only 357k compressed
[17:01:59] <dave-e> hi fenn and everyone else
[17:02:34] <dave-e> just got back from a week of hunting...and trying to ease my reentry into the real world
[17:03:10] <fenn> by escaping to the fantasy-land of yon hinter net?
[17:03:20] <dave-e> indeed
[17:03:38] <dave-e> lots to do
[17:03:43] <dave-e> puppy sounds interesting
[17:03:55] <steves_stallings> puppy-emc can also be run like BDI-live, instructions on download on linuxcnc.org/bdi
[17:04:36] <dave-e> good for testing but I assume slow otherwise
[17:04:59] <steves_stallings> slow boot, runs from ram, but installable to HD
[17:05:34] <dave-e> what drivers come with the iso
[17:06:00] <steves_stallings> you just went beyond my limited knowledge..... 8-(
[17:06:30] <dave-e> guess I'll have to download and look
[17:07:26] <steves_stallings> several people including Ray and Matt are playing with it
[17:07:43] <dave-e> anyone here had any experience with water-cooling cpu's. I'm looking at it not for speed for noise level
[17:08:24] <dave-e> brb .... coffee
[17:12:23] <fenn> mount the motherboard upside down so the water drips _away_ from the expensive electronics
[17:13:05] <anonimasu> fenn: if you fill it with tap water ;)
[17:13:50] <fenn> are there problems with the other chips on the board overheating since there is no air blowing over them?
[17:14:29] <anonimasu> I've never heard about that..
[17:14:32] <anonimasu> but there are low noise fans..
[17:15:22] <dave-e> but not really quiet even then.
[17:15:32] <fenn> i'm wondering about sealing a computer air-tight to keep metal chips out of it
[17:16:00] <fenn> guess i should just try it and see
[17:16:07] <anonimasu> you could circulate the air through a radiator..
[17:16:12] <dave-e> this is on my desktopd...to run cad, sim, etc. in the shop one cannot tell the difference.
[17:17:29] <dave-e> I've also thought about methylene chloride in a sealed system to transfer heat...boils at 40 C...vapor transfer of heat to radiator
[17:17:53] <djb_rh_> djb_rh_ is now known as djb_rh
[17:18:23] <anonimasu> if you want to be sure you can use mineral oil
[17:18:31] <anonimasu> but that requires by far thicker hooses
[17:19:27] <dave-e> the lower viscosity of h20 is much better. one does need to do a good reliable job of plumbing tho
[17:19:48] <anonimasu> yeah
[17:20:03] <anonimasu> but I think there are less visous mineral oil
[17:23:28] <dave-e> typical organics have specific heats in the .45-.55 range ... not good
[17:27:01] <dmess> good mornin' all
[17:28:03] <dave-e> gonna bail...see ya later
[17:28:06] <alex_joni> bon jour
[17:29:42] <dmess> allo alex
[17:30:26] <alex_joni> salut, ca va?
[17:30:58] <dmess> comme ci comme ca... mon pere et a l'hopital
[17:31:45] <djb_rh> holy crap
[17:31:45] <djb_rh> I understood that
[17:32:00] <djb_rh> with merely two years of NC high school French that was 15 years ago
[17:32:08] <dmess> whatts wrong woth that..
[17:32:28] <djb_rh> you obviously don't know much about the quality of our foreign language programs in public schools. :)
[17:32:46] <dmess> oh ok.. ; )
[17:33:03] <dmess> its native to me
[17:33:06] <djb_rh> "je nes parle pas francais" was the only phrase I thought I took from two full years of French
[17:33:27] <djb_rh> since we were required to answer in French any time we were asked anyhting in French
[17:33:46] <dmess> je comprend...
[17:34:00] <djb_rh> heh, I got that, too. :)
[17:34:36] <djb_rh> dmess: so, why is your Dad in the hospital?
[17:34:54] <alex_joni> dmess: sorry to hear that
[17:35:10] <djb_rh> well, it's none of my business...just hope it isn't anything serious
[17:35:12] <dmess> he had a head ache for a few days.. while hunting..
[17:35:19] <djb_rh> ah
[17:35:26] <alex_joni> c'est tres mal?
[17:35:31] <dmess> figured it was the beer .. but it didnt go away...
[17:35:39] <djb_rh> ah
[17:36:03] <dmess> now they've found a spot of blood on his brain 10mm across and 25 mm deep
[17:36:12] <dmess> 100mm
[17:36:16] <alex_joni> ouch
[17:36:25] <djb_rh> yeah
[17:36:40] <djb_rh> I've got a cousin that just had a brain aneurysm at the age of 34
[17:36:55] <djb_rh> they had to do surgery, but she's made a full recovery (this was about six weeks ago)
[17:37:12] <dmess> het not bleedin into his brain yet... just into the gap weteen the skull and the sac that holds the brain
[17:37:48] <dmess> goe under the knive right about now...
[17:38:00] <dmess> going
[17:38:01] <djb_rh> hope it goes well
[17:38:32] <dmess> docs said he was in a best case scenerio
[17:38:40] <rayh> same here. wife has had trouble with aneurisms.
[17:39:01] <dmess> he wasnt shoing any sighns of stroking out at all
[17:41:35] <rayh> That is one good thing. blood inside the brain's sack is a real quick killer/disabler.
[17:43:28] <djb_rh> yeah
[17:43:34] <djb_rh> my cousin was lucky the same way
[17:43:46] <djb_rh> recovery is still *long*, though
[17:44:16] <djb_rh> she started feeling good enough to sit up and eat and even walk around a bit within a couple days, but she got tired *real* fast
[17:44:22] <dmess> doc here seemed to think he'd be home in a few days
[17:44:26] <djb_rh> yeah, she was
[17:44:34] <djb_rh> but couldn't go back to work for at least two months
[17:44:58] <djb_rh> just takes that long to build the stamina back up, apparently
[17:45:12] <dmess> they DO core a hole in your skull...
[17:45:37] <dmess> gonna be some trauma involved
[17:45:49] <djb_rh> yeah
[17:48:38] <dmess> i gotta keep occupied ..going
[17:48:47] <dmess> NUTS here
[17:49:55] <alex_joni> dmess: need smthg to do?
[17:49:58] <alex_joni> * alex_joni grins
[17:51:18] <dmess> yeah... waiting for a phone call... you know...
[17:52:31] <dmess> researching to do another lot of diamonds this evening..
[17:52:44] <rayh> I'll be around the machine but glance in here every so often.
[17:52:53] <alex_joni> hey rayh
[17:52:56] <rayh> rayh is now known as rayh-away
[17:53:01] <alex_joni> wanted to ask you some stuff...
[17:53:07] <alex_joni> but when you come back is ok
[17:53:14] <rayh-away> go ahead.
[17:53:23] <alex_joni> still looking on the tkemc stuff
[17:53:29] <rayh-away> k
[17:53:33] <alex_joni> if I change X Y Z to X Z A in emc.ini
[17:53:45] <alex_joni> then tkemc should still send jog commands for 0,1,2
[17:53:55] <rayh-away> I ran the interpreter with a lathe def X Z only.
[17:53:57] <alex_joni> but it sends 0,2,3 (X=0, Z=2, A=3)
[17:54:11] <alex_joni> interp. should be ok, I think
[17:54:14] <alex_joni> but GUI not
[17:54:15] <rayh-away> Yes it does and it has nothing to do with tkemc.
[17:54:24] <rayh-away> It is the interpreter that does that.
[17:54:29] <alex_joni> how come?
[17:54:41] <alex_joni> tkemc connects to task, not to interp
[17:54:46] <dmess> hard coded axes???
[17:54:50] <rayh-away> Interpreter hard codes x=0 y=1 z=2
[17:54:56] <alex_joni> jog commands don't go through interp.
[17:55:06] <alex_joni> they go from GUI to task, and from there to motion
[17:55:19] <rayh-away> Right. Jog commands are coded to match.
[17:55:29] <alex_joni> let me run some aditional test
[17:55:44] <alex_joni> I tweaked tkemc to see that motion does the necessary jogs
[17:55:53] <rayh-away> cause the motion has to assume a 0 is the first axis,
[17:55:56] <alex_joni> still need to figure out whre the hooks for the readout are done
[17:56:22] <alex_joni> you don't happend to remember that, do you?
[17:56:31] <alex_joni> where the readout on tkemc comes from?
[17:56:57] <rayh-away> Within tkemc you can map em pretty much as you please.
[17:57:13] <rayh-away> for displaying
[17:57:47] <alex_joni> but where do they get mapped?
[17:57:48] <rayh-away> But when you talk to emc through emcsh you must honor EMC's understanding that x=0
[17:57:56] <rayh-away> and so forth.
[17:58:16] <rayh-away> * rayh-away looks for lines
[18:00:11] <rayh-away> In emc/src/emctask/tkemc.tcl they are in lines 41-120 or so.
[18:01:29] <alex_joni> those are the only ones?
[18:01:47] <alex_joni> so worldlabellist and axiscoordmap ?
[18:02:03] <jepler> could you configure emc with axes called X - Z A, then make the front-end skip display of an axis called -?
[18:02:18] <alex_joni> jepler: that's what I am trying
[18:02:24] <alex_joni> and the display works
[18:02:35] <alex_joni> but the problem is to decide what numbering to use on the axes
[18:02:42] <alex_joni> it seems it's not consistent
[18:02:45] <rayh-away> the actual jog commands are issued from the proc at 639
[18:02:51] <rayh-away> proc jogNeg {axis} {
[18:02:54] <alex_joni> motion espects (X=0 Z=1 A=2)
[18:02:55] <rayh-away> or
[18:03:08] <alex_joni> rayh tells me that interp has always (X=0 Y=1 Z=2, etc)
[18:03:15] <rayh-away> proc jogPos {axis}
[18:03:17] <alex_joni> tkemc behaves like that too
[18:03:35] <rayh-away> in line 658
[18:03:37] <jepler> In axis I'm pretty sure we read the axis names from the ini
[18:03:54] <rayh-away> the tickle does that also.
[18:03:56] <alex_joni> jepler: yes, the names (XYZABCRPW)
[18:04:08] <alex_joni> but the problem is: to what numbering do they map?
[18:04:22] <alex_joni> if you have XYZ it's easy (X=0 Y=0 Z=0)
[18:04:46] <jepler> all equal to zero? I don't understand.
[18:04:49] <alex_joni> but if it's XZA then you can have 2 naming conventions (X=0 Z=1 A=2 or X=0 Z=2 A=3)
[18:05:04] <alex_joni> jepler: sorry.. typo (X=0 Y=1 Z=2)
[18:05:21] <alex_joni> thinking way faster than I can type :/
[18:05:46] <dmess> if its a hard coded interpreter.. use the x=0 y=1 a=3
[18:05:50] <jepler> why does the numbering matter?
[18:06:05] <alex_joni> because the NML messages take the number as an argument
[18:06:13] <alex_joni> and if you got that wrong you jog another axis
[18:06:27] <dmess> and it shoud fly it should see the axes you want to display from the ini file .. as i see it
[18:06:49] <alex_joni> say you have (XZA) and you want to jog Z - if you send 1 it might jog Y (which doesn't exist) instead of Z
[18:07:06] <rayh-away> I've found over the years that emc does not see many things the way that I do.
[18:07:17] <alex_joni> jepler: got an axis running around? try setting XZA in the ini, and let me know how jogging works
[18:07:50] <alex_joni> rayh: re jogPos and jogNeg, those are the jogging commands, not the position feedback
[18:08:35] <alex_joni> * alex_joni spots a var alled $position
[18:08:39] <alex_joni> * alex_joni spots a var called $position
[18:08:57] <dmess> actual??? command???
[18:08:57] <rayh-away> damn. my 4.30 emc is borked since the last hour.
[18:09:27] <rayh-away> If you wish position feedback, the loop at the bottom of the file
[18:09:36] <alex_joni> rayh-away: thank you
[18:09:38] <rayh-away> signals emcsh to report back position.
[18:09:50] <alex_joni> dmess: you don't have any difference between actual and commanded in a GUI
[18:09:54] <rayh-away> That position can be abs, rel, machine, and such.
[18:10:44] <dmess> plus or minus the lead /lag.. and you have a differenc between actual and command
[18:10:48] <alex_joni> rayh-away: also, tkbackplot assumes fixed numbering for the axes
[18:11:03] <alex_joni> dmess: that's known to motion, not to GUI
[18:11:08] <rayh-away> yes
[18:11:36] <alex_joni> rayh-away: emc_joint_pos 0..5 ?
[18:11:39] <dmess> and there is nowhere to monitor lead lag thru the GUI??
[18:11:57] <rayh-away> okay emc is running here. Let me illustrate a couple of commanded moves.
[18:12:44] <fenn> all this confusion seems to be over whether you're moving a joint or an axis
[18:12:49] <fenn> xyz are joints
[18:12:53] <fenn> er shit
[18:12:58] <fenn> xyz are axes, 012 are joints
[18:13:01] <alex_joni> not really :D
[18:13:05] <alex_joni> ok.. like that
[18:13:06] <rayh-away> emcTaskPlanExecute(g0 x1 y2 z3) returned 0
[18:13:07] <rayh-away> Outgoing motion id is -2.
[18:13:07] <rayh-away> Issuing EMC_TRAJ_SET_VELOCITY -- (+205,+20, +0,1.247219,)
[18:13:07] <rayh-away> Issuing EMC_TRAJ_LINEAR_MOVE -- (+220,+60, +0,1.000000,2.000000,3.000000,0.000000,0.000000,0.000000,)
[18:13:32] <alex_joni> ok.. wonder what the same is for g0 X1 Z3
[18:13:34] <rayh-away> This is a two axis machine, X and Z are displayed.
[18:13:50] <rayh-away> the g0 x1 y2 z3
[18:13:50] <alex_joni> you can't do XYZ moves on a 2-axes machine
[18:13:54] <alex_joni> that's stupid :/
[18:13:56] <rayh-away> command
[18:14:07] <alex_joni> then it's really borked
[18:14:55] <alex_joni> not sure about emc1, but emc2 (motion controller) doesn't leave any numbers out when numbering joints/axes
[18:14:57] <dmess> so your 2 axis machine WILL take a 3 axis command and NOT give an invaliv address error??
[18:15:02] <alex_joni> so XZ is numbered 01
[18:15:36] <dmess> unless y is present
[18:16:27] <rayh-away> Ah but that is NOT the problem. The problem is that we must create a system of axis numbering that is consistent between emc's interpreter and any sorts of gui stuff sent through or by task.
[18:16:41] <alex_joni> also consistent with the motion controller
[18:17:03] <rayh-away> Every interp motion command has all six positions.
[18:17:17] <rayh-away> exactly like that shown above.
[18:17:43] <fenn> the interp takes AXES as inputs, so that's okay
[18:18:06] <fenn> you dont jog along joints with the interpreter
[18:18:17] <rayh-away> rayh-away is now known as rayh
[18:19:10] <fenn> where is forward kinematics?
[18:19:16] <dmess> sounds like we will need to pad the input with the necessay commas (,)
[18:19:58] <rayh> I guess I missed what you guys are trying to create.
[18:20:11] <dmess> yo dont pick you axes... you omit the ones that arent there
[18:20:29] <fenn> dmess: think about jogging the actuators of a hexapod
[18:20:49] <dmess> yes...
[18:20:59] <rayh> That is exactly what tkemc does. You must have all of the axes defined in the ini up to and through the last one you need.
[18:21:13] <alex_joni_> darn connection
[18:21:15] <alex_joni_> * alex_joni_ is back
[18:21:17] <rayh> and tkemc omits those you do not need.
[18:21:34] <alex_joni_> alex_joni_ is now known as alex_joni
[18:21:36] <steves_stallings> steves_stallings is now known as steves_logging
[18:21:52] <alex_joni> rayh: give me a secodn to read back
[18:21:55] <dmess> but in the process it seems to mis assign axis #'s
[18:21:58] <rayh> k
[18:22:05] <alex_joni> second
[18:22:11] <alex_joni> darn.. can't type today :((
[18:22:26] <dmess> nor i ... shakin too much
[18:23:35] <alex_joni> ok.. read back
[18:23:44] <alex_joni> lets decide on smthg..
[18:23:47] <fenn> does kinematics get linked into motion or task?
[18:23:55] <alex_joni> fenn: wait a bit with kinematics
[18:23:57] <fenn> ok
[18:24:06] <alex_joni> it's not that simple, and if you look at motion you'll figure it out :D
[18:24:09] <dmess> with the hexapod scenerio... your x,y,z axes are ALWAYS going to go thru the interprter... just due to the kinematics of the system..
[18:24:12] <alex_joni> you have 3 cases
[18:24:29] <alex_joni> KINEMATICS_IDENTITY (e.g. joints=axes)
[18:24:33] <dmess> ita always an interpolated move
[18:24:35] <alex_joni> KINEMATICS_FORWARD
[18:24:40] <alex_joni> KINEMATICS_BOTH
[18:24:52] <alex_joni> when you talk hexapod you're in case 3
[18:24:58] <alex_joni> and you jog using TELEOP commands
[18:25:06] <alex_joni> if you want to jog on carthesian space
[18:25:20] <alex_joni> or you switch to joint view, and jog each joint one by one
[18:25:47] <fenn> ok, so why isn't it always like that?
[18:25:49] <alex_joni> but in that case you lose youre homing, and need to rehome before switching back to world view
[18:26:10] <alex_joni> fenn: don't want to run unnaccessary things (kinematics calcs in case 1)
[18:26:16] <alex_joni> and not applicable in case 2
[18:26:22] <alex_joni> also.. TELEOP jog is strange
[18:26:31] <alex_joni> you can jog only with a specified speed
[18:26:42] <alex_joni> can't do exact jogs (e.g. incremental)
[18:26:49] <fenn> surely performance isn't an issue when jogging though
[18:27:21] <alex_joni> nope.. but if you decide to do case 3 always, you'll do that all the time
[18:27:28] <alex_joni> not only when jogging
[18:27:39] <alex_joni> jogging is actually a special case of movement commands
[18:27:49] <dmess> how much overhead will it cost??
[18:27:56] <alex_joni> dmess: don't care
[18:28:02] <alex_joni> it's not going to happen
[18:28:05] <alex_joni> doesn't make sense
[18:28:06] <dmess> so DO it
[18:28:15] <alex_joni> you DO it, if you want
[18:28:23] <alex_joni> lets go back to what the problem is
[18:28:42] <alex_joni> rayh: still around?
[18:29:41] <alex_joni> see.. now you scared him away
[18:30:10] <dmess> i see you case 1 as a milling machine or lathe... no need for real kinematics unless you have stupid nutating heads or tables...
[18:30:27] <dmess> case 2... special cases....
[18:30:29] <alex_joni> right.. default is case 1
[18:30:42] <alex_joni> easiest case
[18:30:43] <alex_joni> and almost in all cases applied
[18:30:47] <alex_joni> very few cases need 3
[18:30:51] <dmess> case 3 is for hex aposd and other motion control apps
[18:31:00] <alex_joni> and even less ones do 2
[18:31:17] <dmess> precisely..
[18:31:18] <alex_joni> but 2 is stupid, because you don't have inverse kins
[18:31:30] <alex_joni> so let's make 1 work how it should
[18:31:33] <dmess> but it would be nice
[18:31:49] <alex_joni> and not spend time on talking about 2 and 3 (which btw DO work as they should)
[18:32:05] <alex_joni> * alex_joni ran one machine with 3 beautifully
[18:32:24] <fenn> your "ufo"
[18:32:28] <alex_joni> right
[18:32:40] <dmess> the initial issue with 1 was that JOG doesnt go thru the interp... so if you set xza
[18:32:41] <alex_joni> ok.. the problem as I see it is in naming conventions
[18:32:57] <dmess> thins wanted to go screwy
[18:33:06] <alex_joni> not really
[18:33:09] <alex_joni> the problem is this:
[18:33:23] <alex_joni> different components use different numbering on the axes
[18:33:37] <alex_joni> and when they talk to each other surely it won't work as it should
[18:33:44] <alex_joni> imagine this (which is the case):
[18:33:59] <alex_joni> motion counts the axes from 0, without leaving any unused
[18:34:19] <alex_joni> so if you have X,Z,C they get counted as 0,1,2
[18:34:27] <alex_joni> inside the motion controller
[18:34:43] <alex_joni> now.. the commands to move an axis (or joint) have a number in there (which one to move)
[18:35:05] <alex_joni> the interp (rayh said this, need to verify it) have fixed numbering
[18:35:23] <alex_joni> so X,Z,C get numbered as 0,2,6
[18:35:33] <alex_joni> same does tkemc iirc
[18:35:34] <dmess> yes he did they are hard coded
[18:35:54] <alex_joni> ok.. imagine a jog on axis Z now..
[18:36:02] <fenn> wait a minute, you can't jog on an axis
[18:36:03] <alex_joni> tkemc sends Jog axis #2
[18:36:10] <alex_joni> fenn: why not?
[18:36:19] <fenn> because an axis only exists in your mind
[18:36:25] <dmess> case 1
[18:36:49] <fenn> the way i see it, XYZRPW will always be there no matter what
[18:37:02] <fenn> cartesian coordinates are not going to change whether your machine has them or not
[18:37:23] <dmess> and the physical thing you WaNT to move is the axis
[18:37:30] <fenn> dmess: no, that's a joint
[18:37:36] <fenn> it's always a joint
[18:37:49] <dmess> and a lead screw is a rotary joint
[18:37:51] <fenn> am i making sense?
[18:37:54] <alex_joni> fenn: you're saying I can't jog on axis 2 ?
[18:37:57] <alex_joni> fenn: you're saying I can't jog on axis Y ?
[18:38:03] <alex_joni> sorry bout the first one?
[18:38:06] <alex_joni> that's stupid
[18:38:17] <alex_joni> my machine has XYZ axes
[18:38:26] <alex_joni> I want to jog on axis Y
[18:38:29] <dmess> read the 3 case scenerios alex presented
[18:38:32] <fenn> ok
[18:38:39] <alex_joni> I agree that in this special case it coresponds to join #2
[18:38:44] <alex_joni> but that's not always true
[18:39:01] <fenn> right, doesn't it have to check which type of kinematics are in use first?
[18:39:18] <dmess> depends on whether emc or motion controller is seeing it
[18:39:36] <alex_joni> the motion controller does that
[18:39:57] <alex_joni> but like I said (there are 2 different JOG commands for different kins)
[18:40:22] <alex_joni> on KINEMATICS_BOTH and KINEMATICS_FORWARD you don't jog with normal JOG commands, you send TELEOP_JOG commands
[18:40:29] <dmess> one is kins off.... one is kins on ( completely )
[18:40:32] <alex_joni> but back to what I was saying
[18:40:58] <alex_joni> dmess: kins are always on, but in case 1 they just do joint0=axisX
[18:41:19] <dmess> i understand... im keeping it simple..
[18:41:28] <alex_joni> ok..
[18:41:53] <alex_joni> so back to where we were
[18:41:58] <alex_joni> if I want to jog Z
[18:42:07] <alex_joni> I need to send a JOG commands (we are in case 1)
[18:42:20] <alex_joni> and along with that jog command, I need to specify the axis number
[18:42:49] <fenn> so, since you are in case 1, why would your numbering scheme be unconventional?
[18:43:00] <alex_joni> because I don't have all axes
[18:43:07] <alex_joni> I don't have XYZ, but XZA
[18:43:10] <alex_joni> Y is missing
[18:43:12] <dmess> you skipped 1 or 2
[18:43:25] <alex_joni> now the problem do I count 023? or 012?
[18:43:28] <fenn> can't you just make it number them conventionally?
[18:43:33] <fenn> numbered 023
[18:43:36] <alex_joni> what's conventionally?
[18:43:39] <alex_joni> who says that?
[18:43:45] <fenn> the interp
[18:43:53] <alex_joni> what if you have XZAP ?
[18:44:02] <alex_joni> not P, R
[18:44:02] <fenn> the "rs274 standard" way of doing things is conventional
[18:44:29] <fenn> uh, i dont know the difference between C and R
[18:44:34] <alex_joni> point me out where this is specified in the rs274 standard
[18:44:39] <fenn> it's in the code
[18:44:46] <fenn> the interp is your holy artifact
[18:44:50] <dmess> or the GUi has to pas x,,z,a,,,,,p for every motion command
[18:44:59] <alex_joni> fenn: not really
[18:45:13] <alex_joni> dmess: that would involve changing the NML messages
[18:45:52] <dmess> sorry..
[18:45:54] <fenn> alex_joni: what about this for an edge-case:
[18:46:03] <fenn> you have XZABCRPW
[18:46:11] <alex_joni> you can't
[18:46:12] <fenn> (can emc even handle that?)
[18:46:15] <alex_joni> 6 is the max
[18:46:27] <fenn> what about 6 rotary axes
[18:46:42] <alex_joni> you need XYZABC
[18:46:46] <alex_joni> that's what emc can handle
[18:46:53] <dmess> good for coming bac to where youve been i guess
[18:47:24] <Jymmm> This is VERY cool... http://www.patmedia.net/marklevinson/cool/cool_illusion.html
[18:47:28] <dmess> ther are only rotary axes guys...
[18:49:39] <fenn> since you can only have XYZABC, what is the problem with numbering them 0123456?
[18:49:46] <fenn> heh
[18:49:48] <fenn> 012345
[18:49:57] <alex_joni> no problem
[18:50:03] <alex_joni> absolutely no problem
[18:50:11] <alex_joni> but if one is missing you have a problem
[18:50:16] <fenn> why?
[18:50:16] <dmess> buy with which names assigned and in what order
[18:50:22] <alex_joni> XZABC should that be 02345 ?
[18:50:29] <fenn> yep
[18:50:32] <alex_joni> or 01234 as the motion controller does it?
[18:50:43] <fenn> why would you do 01234?
[18:50:54] <dmess> but the interp is hard coded otherwise guys
[18:50:55] <alex_joni> because that's the way it's done
[18:51:11] <fenn> so can we just change it?
[18:51:17] <dmess> no
[18:51:23] <alex_joni> why not?
[18:51:37] <dmess> legacy..
[18:51:51] <alex_joni> it's not part of the interpreter anyways
[18:51:54] <fenn> but it doesnt work anyway, so fixing it won't break anything
[18:52:03] <alex_joni> interpreter calles canonical functions, so it's done there
[18:52:20] <alex_joni> interpreter only knows XYZABC
[18:52:23] <alex_joni> no joints or numbers
[18:53:23] <dmess> ok so we use pseudo names for those axes if we want then different and problem solved
[18:53:46] <dmess> them
[18:55:21] <dmess> its up to the GUI and the command interpreter to jive with the motion controllers...whhich will be 012345
[18:55:52] <dmess> order up a fanuc or siemens drive...
[19:00:37] <fenn> is this where it names the axes?:
[19:00:39] <fenn> /* read and process per-joint inputs */
[19:00:39] <fenn> for (joint_num = 0; joint_num < EMCMOT_MAX_AXIS; joint_num++) {
[19:01:00] <fenn> bleh that didnt mean anything
[19:01:04] <fenn> control.c line 286?
[19:01:58] <alex_joni> alex_joni is now known as alex_joni_busy
[19:02:18] <dmess> if you set 3 there it mot controll will expect 012
[19:02:34] <dmess> not 023
[19:03:21] <fenn> i guess alex has a real job after all
[19:03:54] <dmess> gotta pay the bills somehow
[19:04:25] <fenn> okay that was obviously the wrong line of code
[19:04:37] <fenn> * fenn digs some more
[19:05:06] <dmess> its not about what YOU call your axes... its about what motion is hard codded to look for
[19:05:45] <dmess> use xyz but program yzx and see if the program works
[19:06:33] <dmess> sorry.. xy a... would work
[19:06:52] <dmess> xy b ,c, , etc.... would work
[19:07:29] <dmess> but set yza... and program xyz
[19:07:48] <dmess> and it gets ugly
[19:08:30] <fenn> dmess: it doesn't pass the arguments that way
[19:08:40] <fenn> it passes them one at a time
[19:09:05] <fenn> each nml "jog" message has one and only one axis
[19:09:10] <dmess> still going to the wrong address
[19:13:20] <dmess> look up some old siemens 300m documentation... they were used on universal machines... so a g81 drill code HAD to have an axis # to drill on
[19:17:40] <jepler> in terms of emc's g-code, applying the g17/g18/g19 "plane selection" to canned cycles might make sense
[19:18:21] <jepler> though a new g-code would need to be introduced to avoid chainging the meaning of existing code which programs g18 or g19 and uses canned cycles
[19:18:43] <jepler> oh, wait, that's what they do already.
[19:18:53] <jepler> http://www.linuxcnc.org/handbook/RS274NGC_3/RS274NGC_33a.html#1003380
[19:19:09] <dmess> try it... tell me if it works... ive used those codes for going on 20 yrs and still have to blow bubbles out my ass to use an angle head
[19:20:58] <jepler> I only know what the documentation says
[19:21:07] <jepler> (and it still wouldn't help if you were moving at an angle)
[19:29:57] <dmess> i have my APT software to give me xyz's for any compound angle and depth of hole now... no cycles involved
[19:35:31] <anonimasu> wb alex_joni
[19:35:54] <alex_joni> ty
[19:36:56] <Jacky^> Jacky^ is now known as Jacky^afk
[19:37:37] <dmess> who want to play with some software to make it work on linux??
[19:39:31] <anonimasu> make it work?
[19:42:28] <dmess> works in an old dos system...
[19:42:46] <anonimasu> have you tried running dosemu?
[19:42:48] <anonimasu> what is it?
[19:42:52] <Jacky^afk> how much the fish price ?
[19:42:58] <dmess> can you handle a zip file... ; )
[19:43:02] <Jacky^afk> ops wrong window
[19:43:06] <Jacky^afk> :P
[19:43:52] <dmess> APT Automatically Programmed Tool.... developed by nasa in the '60s
[19:44:40] <anonimasu> hm I guess you could run it with dosemu
[19:44:58] <Jymmm> * Jymmm shoves Jacky^afk into #seafood
[19:45:30] <dmess> it was pretty picky as to want dos 6.22 to post... but i guess we could build a linux post processor..
[19:45:48] <anonimasu> dmess: I could have a look at it
[19:46:34] <dmess> how should i send it to you
[19:46:57] <dmess> shit my dad just NOW went into surgery
[19:47:17] <anonimasu> mail
[19:48:19] <dmess> mail isnt set up on this box.. addy??
[19:49:02] <anonimasu> I sent you a message with it
[19:49:02] <anonimasu> :)
[19:50:07] <anonimasu> or if you can dcc it..
[19:53:56] <dmess> i havent found the dcc option on iirc yet.. LOL
[19:55:29] <dmess> check you mail... its a verbose language that allows you to define EXPLICIT geometry and drive a tool along it... by
[19:56:02] <dmess> defining the tool and iteratively. calculating where it should be
[19:56:21] <dmess> and sending it there
[19:57:15] <dmess> but its VERY fortran'ish
[19:58:40] <dmess> you have it yet??
[20:01:24] <dmess> should be a test.apt file there ... does a square 100mm x 100mm
[20:02:09] <dmess> planes and all... cylinders .. i even had to define a torus once...
[20:02:44] <dmess> and drive a tool along the inside...
[20:03:20] <dmess> .. like here we need to cut the inside of this donut...
[20:04:06] <dmess> 26 lines of code later... theres the g-code... wanna RACE
[20:07:29] <etla> hi
[20:12:04] <Jacky^afk> Jacky^afk is now known as Jacky^
[20:13:09] <anonimasu> let me see
[20:13:19] <anonimasu> my linux crapped out
[20:13:20] <anonimasu> heh
[20:13:40] <anonimasu> the network card refused to talk to my network
[20:22:05] <alex_joni> wrong network
[20:22:44] <Jymmm> wrong oeperator
[20:24:03] <Jymmm> Please hanf up and try your call again.
[20:26:56] <anonimasu> heh
[20:26:56] <anonimasu> yeah
[20:39:31] <djb_rh> NO CARRIER
[20:40:18] <Jymmm> at
[20:40:22] <Jymmm> CONNECT 9600
[20:40:58] <Jymmm> atdt900sexyray
[20:41:04] <Jymmm> CONNECT 69
[20:52:57] <etla> i cen't believe some people are still using dialup....
[20:53:24] <Jymmm> Some people have no choice
[20:53:34] <Jymmm> rayh happens to be one of them.
[20:56:00] <dmess> ahh good old copper lines are secure thouth
[21:00:21] <alex_joni> * alex_joni read a lot of code...
[21:00:27] <alex_joni> anyone interested in conclusions?
[21:00:32] <anonimasu> sure
[21:00:38] <alex_joni> ok.. it's like this:
[21:00:50] <alex_joni> axes are called XYZABC, and numbered 012345
[21:00:59] <Jymmm> * Jymmm excersies... running out at the mouth, jumpin to conclusions, etc.
[21:01:04] <Jymmm> s/out/off/
[21:01:07] <alex_joni> if any is missing then the numbers are missing (XZA -> 023)
[21:01:16] <alex_joni> so far so good
[21:01:29] <alex_joni> but.. there are a lot of places where axes are confused with joints
[21:01:49] <alex_joni> so in that case (axes 023) we really should have joints (012)
[21:01:55] <alex_joni> and never mix the two
[21:02:02] <fenn> that's what i was saying
[21:02:37] <fenn> where does motion number the axes?
[21:03:23] <alex_joni> it doesn't
[21:03:27] <alex_joni> it numbers the joints
[21:03:31] <alex_joni> and then calls kins
[21:03:33] <fenn> in trivkins.c right?
[21:03:44] <alex_joni> to get the position in axes
[21:03:47] <fenn> joints[0] = pos->tran.x; etc
[21:03:59] <alex_joni> that's the convertion from axes to joints
[21:04:13] <alex_joni> and in the same file you have the conversion from joints to axes
[21:04:24] <alex_joni> pos->tran.x = joints[0]
[21:04:37] <alex_joni> but ... when I see code like this:
[21:04:55] <alex_joni> emc_axis_activate (int axis)
[21:05:05] <alex_joni> which sends EMCMOT_ACTIVATE_JOINT
[21:05:07] <fenn> yeah that makes no sense
[21:05:08] <alex_joni> smthg is fishy
[21:05:26] <fenn> this is the result of no type checking
[21:05:37] <alex_joni> not really type checking
[21:05:55] <alex_joni> let me try smthg
[21:06:03] <fenn> you should have to pass a joint object to EMCMOT_ACTIVATE_JOINT
[21:06:12] <fenn> er, i got that wrong
[21:06:23] <fenn> it's all screwed that's the problem :)
[21:06:52] <djb_rh_> well, I understood dmess's french earlier, but it took me forever to get that alex_joni means "something" when he says "smthg"
[21:07:00] <djb_rh_> :)
[21:07:05] <alex_joni> djb_rh_: LOL
[21:07:08] <alex_joni> did you get that?
[21:07:28] <fenn> i found myself reading indonesian the other day
[21:07:38] <fenn> it was kinda unexpected
[21:07:58] <alex_joni> * alex_joni found himself reading/writing tcl today
[21:08:34] <Jymmm> alex_joni you sick bastard!
[21:08:50] <alex_joni> well.. I did fix the jogging bug
[21:09:01] <alex_joni> the one which allows you to jog multiple axes at one time
[21:09:14] <fenn> without pressing any keys
[21:09:27] <Jymmm> mind control joggin huh?
[21:09:48] <fenn> yeah it's a feature
[21:10:00] <Jymmm> well shit, if we have mind control cnc, what do we need emc for?
[21:10:25] <alex_joni> to control our mind
[21:16:27] <alex_joni> bbl
[21:56:12] <dmess> time . by picnk floyd.... CRANKIN...
[21:58:15] <dmess> both big speakes still work as long as the amp cranked up...
[21:59:06] <rcsu> bad amplifier :)
[21:59:56] <dmess> dirty knobs and pots
[22:00:24] <rcsu> change em, its worth the efford if the rest is good
[22:00:28] <fenn> "As a general rule, don't solve puzzles that open portals to Hell. "
[22:00:46] <rcsu> ?
[22:00:56] <fenn> nothin
[22:00:57] <dmess> they need a cleanin.. is all..
[22:01:12] <rcsu> oh, open potis ?
[22:01:41] <dmess> yes.. audio amp
[22:02:12] <dmess> archieic DYNACOUSTICS
[22:02:39] <dmess> but still drives what i got
[22:02:46] <rcsu> mine has closed
[22:02:54] <rcsu> dunno that
[22:03:52] <dmess> even closed i open them.. sorry i havent listened since i was akid and havenet learned
[22:04:39] <dan_falck> dmess: tell me more about the APT program that you have
[22:04:49] <dan_falck> any source code?
[22:05:00] <dmess> apt source??
[22:05:10] <dan_falck> for the apt engine
[22:05:39] <dmess> no source for that ... i wish.
[22:05:46] <dan_falck> ok, just hoping
[22:06:00] <dan_falck> I almost bought some of the source 5 years ago
[22:06:09] <dan_falck> but it was only on a VAX tape
[22:06:12] <dmess> just hoping someone could make this one work
[22:06:25] <dan_falck> that I didn't have any way of exporting to something modern
[22:06:30] <dmess> no IBM 360 too
[22:06:39] <dan_falck> $1200 from CAM-I
[22:06:53] <dmess> who are they
[22:07:04] <dmess> not this stuff..
[22:07:32] <dan_falck> a US gov't related group that is supposed to help industry with stuff like APT
[22:07:41] <dan_falck> tech transfer outfit
[22:07:47] <dan_falck> based in Texas
[22:08:10] <dmess> really. since when.. i gave up lookin about 2 yrs ago..
[22:08:27] <fenn> anyone here heard of freesteel?
[22:08:44] <dan_falck> I talked to one of them about buying the code, but they didn't want me to open it to the public
[22:08:52] <dmess> nope i think i know where your heading....
[22:08:53] <anonimasu> dan_falck: is the engine really that good?
[22:09:17] <dan_falck> it was the basis for a lot of the modern CAM systems
[22:09:39] <fenn> i dont understand what the big difference is
[22:09:53] <dmess> i never got the 1200$$ option but i remember a thing about NOT FOR EXPORT... im in Canada
[22:10:01] <fenn> it uses the side of the cutter, big deal. why is that news?
[22:10:06] <dmess> the other blue state
[22:10:49] <dan_falck> which group did you talk to (dmess)
[22:11:39] <dmess> catia uses dumb apt to post to EVERYTHING.... for 1200$$ us get the code... I WILL PAY IT
[22:12:23] <anonimasu> is catia good ;)*grins*
[22:12:29] <dan_falck> I will look into it again- if they have a paper print out
[22:12:29] <Jymmm> * Jymmm waves to paul_c
[22:12:37] <dmess> its the ASSOCIATIVITY....
[22:12:46] <dmess> not goto's..
[22:12:51] <anonimasu> I've used catia..
[22:13:03] <dmess> got intof lx and cyl3
[22:13:16] <anonimasu> but it's bulky compared to solidworks :)
[22:13:51] <dmess> it HAS a mathematical algorithm in it and an arlem in it
[22:14:22] <dmess> catia v5 IS solidwirks with nc
[22:14:45] <dmess> dassault bought it..
[22:15:02] <anonimasu> the catia I ran were so packed with features that I got scared of it..
[22:15:13] <anonimasu> ;)
[22:15:39] <dmess> me too.... v 4 r 15 was baptism.. on a unx box
[22:16:35] <dmess> 3 knobs and a box with 40 buttons... for 1 hand... the other held the mouse
[22:18:33] <dmess> 3 button with and if or capability
[22:23:01] <dmess> its all about updates..
[22:26:16] <anonimasu> updates?
[22:30:03] <dmess> engineering changes.. update the geo... the toolpath tacke care of itself...
[22:30:31] <anonimasu> yep
[22:30:34] <dmess> unless its a major change...
[22:30:59] <dmess> which require NEW geo
[22:31:47] <dmess> i did the upper and lowe torsoin links for the 777 in verbose script
[22:32:14] <dmess> from raw Ti forgings
[22:35:20] <dmess> anon.. you have the apt code
[22:35:39] <anonimasu> yeah
[22:36:06] <dmess> sorry.. pre dispositiond to have brair farts
[22:37:06] <dmess> whos still has a 486
[22:38:26] <dmess> ive got bones that could reassembl one.... "could we all post from the same box globally??"
[22:39:12] <dmess> bcz the apt works on 98 boxes
[22:39:24] <dmess> cant post though
[22:40:50] <alex_joni_busy> alex_joni_busy is now known as alex_joni
[22:42:02] <paul_c> Hey Alex.... You put fenn straight on axis naming conventions ?
[22:42:16] <alex_joni> kinda..
[22:42:17] <fenn> not really
[22:42:24] <alex_joni> but I think something is still borked
[22:42:34] <alex_joni> not sure about emc1 but I guess it is there too
[22:42:38] <dmess> APT is iterative... and therefore takes cpu power
[22:43:12] <dmess> yes i see borked too..
[22:45:47] <dan_falck> dmess: what's your email address?
[22:46:49] <dmess> mes_for_fun@yahoo.com... will do
[22:47:06] <dmess> should i ask why??
[22:47:34] <dmess> and its ALL about fun...
[22:47:44] <dan_falck> APT stuff
[22:48:02] <dmess> send along..
[22:48:31] <dmess> brb dad check up
[22:48:40] <anonimasu> I dont quite get how you would machine a torso with it..
[22:48:46] <anonimasu> or well inside one
[22:52:03] <anonimasu> wb alex
[22:52:10] <alex_joni_> darn connection
[22:52:16] <alex_joni_> looked fine, but I was out
[22:52:39] <alex_joni_> time to kill
[22:52:41] <alex_joni_> time to die
[22:52:58] <alex_joni_> alex_joni_ is now known as alex_joni
[23:11:01] <dmess> i just talked to my dad he seems fine.. remembers yesterday's moose phone call.. and he seems to be himself on meds
[23:11:56] <alex_joni> dmess: nice to hear that
[23:11:59] <alex_joni> wish him all the best
[23:12:42] <dmess> thnx all
[23:13:28] <dmess> he's the thoughest dad ive ever had....
[23:13:44] <alex_joni> hopefully the only one you'll ever have
[23:14:55] <dmess> and only..
[23:15:53] <dmess> but BLUE TWISTED STEEL TOUGH>>> went to hell... a few times in different holes
[23:16:54] <dmess> 8370' below collar...7390.... 6490.
[23:21:30] <dmess> he sounds GOOd . im extatic..... dads gonna be okay.... i m smilin... ; )
[23:23:48] <alex_joni> nice
[23:53:23] <fenn> why are cad programs so huge?
[23:53:54] <alex_joni> because they are expensive
[23:54:53] <fenn> "free-cad-libs" is 69 megs and is taking forever