#emc-devel | Logs for 2007-07-03

[02:34:05] <jepler> wow, what a clusterfuck gcode is. we should stop supporting it.
[02:34:09] <jepler> * jepler looks at the discussion on #emc
[02:34:33] <cradek> yeah, yuck
[02:34:49] <Skullworks-PGAB> * Skullworks-PGAB giggles
[02:37:20] <cradek> jepler: I'm bravely soldiering through adding uvw motion (I intend to put it on a branch)
[02:40:16] <cradek> jepler: I have no idea where (in the var file) to save the new offsets
[02:40:52] <cradek> I suppose I'll expand it and push everything down, but ... yuck
[02:41:33] <Skullworks-PGAB> cradek - could you just pick up at 6xxxx
[02:41:57] <Skullworks-PGAB> like the g55 address +1000
[02:42:25] <cradek> Skullworks-PGAB: I'm not sure which is worse
[02:43:06] <Skullworks-PGAB> Haas did a chop job like that for awhile
[02:43:29] <Skullworks-PGAB> went g59 then jumped to like G140
[02:49:34] <jepler> cradek: isn't there room? the current ones run from 5xx1 to 5xx6
[02:49:48] <jepler> so you have 789 available
[02:49:56] <cradek> oh??
[02:50:08] <jepler> that's what the documentation says
[02:50:10] <cradek> (I didn't even look yet)
[02:50:28] <jepler> actually there's a whole lot available, they are spaced 20 apart
[02:50:42] <cradek> url?
[02:51:02] <jepler> http://linuxcnc.org/docs/2.1/html/gcode/main/#998
[02:51:36] <cradek> g92 is close, but still ok
[02:51:56] <cradek> well that's a relief
[02:52:06] <cradek> thanks, mists of time
[10:13:05] <anonimasu> :)
[10:13:20] <alex_joni> :) ?
[10:13:50] <anonimasu> 04:36 <@jepler> wow, what a clusterfuck gcode is. we should stop supporting it.
[10:34:50] <alex_joni> heh
[10:39:18] <anonimasu> alex_joni: awake?
[10:40:06] <anonimasu> alex_joni: is the UWV support for variaxis style machines?
[10:40:31] <alex_joni> anonimasu: for now it'll be only g-code support
[10:40:35] <anonimasu> * anonimasu nods
[10:40:43] <alex_joni> afterwards we need to see how it can be used
[10:40:45] <anonimasu> I hope we will get there :p
[10:40:52] <anonimasu> so we can do 5 axis contouring ;)
[10:41:08] <anonimasu> though it'll be a while before anyone has a machine for it I guess..
[13:29:24] <Guest827> Guest827 is now known as skunkworks_
[15:21:05] <jepler> hi skunkworks_
[15:21:23] <jepler> you didn't do any testing of index pulse this weekend, right?
[15:29:17] <skunkworks_> No - sorry. Goofed off all weekend.
[15:53:11] <skunkworks_> just for reference - I jogged a long distance (50inches) and with the ff0 set to -10 - the following error stayed the same. (zeroed out)
[17:10:10] <anonimasu> hm
[17:10:16] <anonimasu> do you need to use FF on a usc machine=
[17:10:18] <anonimasu> ?
[17:10:25] <anonimasu> err tune the pid loops?
[17:10:27] <anonimasu> even if you run without feedback?
[17:13:52] <jepler> anonimasu: yes
[17:14:05] <jepler> anonimasu: the "feedback" from the USC is the actual number of steps issued
[17:14:21] <anonimasu> jepler: maybe I should tune it..
[17:14:36] <anonimasu> jepler: maybe my machine would work bette r:)
[17:15:03] <anonimasu> or I should just hook up this gecko feedback thing..
[17:15:09] <anonimasu> or get real servodrives.
[17:19:22] <jepler> it would be nice if the usc had an internal, automatically tuned loop so that you could command it with position instead of velocity
[17:19:35] <jepler> but it doesn't, and won't until someone contributes it
[17:20:36] <anonimasu> yeah
[17:23:55] <anonimasu> hm
[17:24:02] <anonimasu> * anonimasu just started with the gui stuff
[17:27:08] <anonimasu> it's pretty easy if I dont need to care about a preview..
[17:38:46] <skunkworks_> must be about 12 years old
[17:39:16] <anonimasu> skunkworks_: what?
[17:39:22] <anonimasu> that guy?
[17:39:39] <skunkworks_> :)
[17:39:51] <jmkasunich_> naw - he
[17:40:04] <jmkasunich_> he's 80 and arthritic, so he needs as many typing shortcuts as possible
[17:40:24] <skunkworks_> Hi jmkasunich_: Took today off?
[17:40:31] <jmkasunich_> yeah
[17:40:34] <skunkworks_> Nice
[17:40:33] <jmkasunich_> plumber
[17:40:37] <jmkasunich_> not nice
[17:40:39] <skunkworks_> eww
[17:40:53] <jmkasunich_> $$$ not nice
[17:41:00] <jmkasunich_> or should that be $$$$$$$$$
[17:41:09] <jmkasunich_> (upcoming, just found out about the need today)
[17:41:16] <skunkworks_> that is some of the stuff I don't mind doing my self. that and electrical.
[17:41:30] <skunkworks_> I just don't like trim and finish work
[17:41:41] <jmkasunich_> broken sewer pipe 8' underground isn't a do-it-yourself thing
[17:41:56] <skunkworks_> it is when you have a back hoe ;)
[17:42:01] <skunkworks_> but yes. yeck
[17:42:13] <jmkasunich_> 8' underground under the front porch ;-(
[17:42:26] <skunkworks_> ouch.
[17:43:18] <skunkworks_> the lateral line was replaced in the house we just bought - it had caved in. THey had put in a new basement - but only connected to the old line to the main sewer.
[17:43:51] <jmkasunich_> so they were running all the wastewater into the wrong (caved in) pipe?
[17:46:58] <skunkworks_> It caved in a few months after the house was moved.
[17:47:12] <skunkworks_> But I am sure it was already leaking.
[17:47:15] <skunkworks_> old neighborhood
[17:47:40] <skunkworks_> so we got a new sidewalk also :)
[17:48:54] <skunkworks_> jmkasunich_: don't go to the bathroom until they are done ;)
[17:49:03] <jmkasunich_> heh
[17:49:13] <anonimasu> :D
[17:49:22] <jmkasunich_> this has been a not-fun day
[17:50:10] <jmkasunich_> we've had problems every so often due to roots - call the county, they come out and snake from the cleanout in the front lawn to the main sewer pipe in the street, and its fine again (for a couple years)
[17:50:27] <jmkasunich_> I expected today to be just another one of those days
[17:50:34] <jmkasunich_> but they didn'
[17:50:48] <jmkasunich_> didn't find much, and when they left, the drain was still very slow
[17:50:53] <skunkworks_> Wouldn't unplug - or did they look at it with a camera and find out there was nothing left/
[17:51:05] <jmkasunich_> the county guys are lazy (but free)
[17:51:14] <skunkworks_> heh
[17:51:23] <jmkasunich_> no camera or anything, 10 mins with the snake and they're outta here
[17:51:30] <skunkworks_> ah
[17:51:33] <jmkasunich_> so I called a real plumber
[17:51:53] <jmkasunich_> who got it running again (after quite a bit of snake work)
[17:52:08] <jmkasunich_> but I asked them to camera it, and when they did, ouch
[17:52:27] <jmkasunich_> note: you never want to see the outside of a sewer pipe from one of those cameras
[17:52:35] <skunkworks_> I bet not.
[17:52:41] <jmkasunich_> "how did it get out there?"
[17:52:58] <skunkworks_> was it drain tile or cast pipe?
[17:53:04] <jmkasunich_> tile
[17:53:06] <skunkworks_> eww
[17:53:41] <skunkworks_> good to replace it then.. won't be a problem in your life time most likely. Just hope they get the pitch right.
[17:54:05] <jmkasunich_> its the digging that will suck
[17:54:22] <jmkasunich_> its in no-mans land
[17:54:35] <skunkworks_> how far by hand?
[17:55:05] <jmkasunich_> the house was originally a bungalow, basement wall = front wall, but sometime they added a full width front addition
[17:55:27] <jmkasunich_> 2/3 is porch, and 1/3 is a foyer (interior space)
[17:55:52] <jmkasunich_> the pipe seems to be just below the midpoint of the wall between foyer and porch (under the front door basically)
[17:56:38] <skunkworks_> that is where the cast ends and the tile starts?
[17:56:47] <jmkasunich_> dunno
[17:57:00] <jmkasunich_> phone
[18:08:33] <skunkworks_> jmkasunich_: when you get a chance will you look at these pictures.
[18:08:32] <skunkworks_> http://www.electronicsam.com/images/KandT/servostart/tune/
[18:19:33] <skunkworks_> jmkasunich_: some back ground. when I get PID tuned 'ok' I have the .0008 steady state error at 100ipm. WHich jepler pointed out is how far it moves during one servo cycle. .5ms. Now I am told that ff1 should fix that. but as you can see - ff1 as ff2 seems to only effect the accel/decel areas. The only thing that seems to effect the steady state error is ff0 - which I am told is wrong as I understand it - the further
[18:31:24] <jmkasunich_> skunkworks_ have you looked at the following error while moving BOTH directions?
[18:31:35] <jmkasunich_> I would expect FF0 to have an assymetrical effect
[18:31:53] <jmkasunich_> if it improves ferror jogging +X, I bet it will hurt it going -X
[18:33:25] <skunkworks_> I actaully tried it both directions - and if I zero the ferror with ff0 - it does seem symetrical both directions.
[18:34:15] <jmkasunich_> strange
[18:34:29] <skunkworks_> :) yes - people think I am crazy.
[18:34:43] <jmkasunich_> think?
[18:35:00] <skunkworks_> I checked that everything was hooked up correctly - all the ff0-ff2 are hooked up to the right pins.
[18:35:34] <skunkworks_> I even did a 'show' after changing a few of the parameters to make sure they where changing the right pins.
[18:35:51] <jmkasunich_> you mean the setp pid.0.ff0 [AXIS_0]FF) lines in the hal file?
[18:35:58] <skunkworks_> yes
[18:36:08] <jmkasunich_> those aren't pins, they're params, but its good that you've checked that
[18:36:32] <jmkasunich_> I just checked the source, ff0 is multiploed by the commanded position and added to the toutput
[18:36:45] <jmkasunich_> that means that its effect will be different at different places on the machine
[18:37:05] <skunkworks_> That is what I understand also from what cradek had said.
[18:37:14] <jmkasunich_> I can't explain why it help you in your tests, but I really don't think you want to use it
[18:37:32] <jmkasunich_> did you say you were getting within 0.0005 at 100ipm without it?
[18:37:59] <skunkworks_> the steady state error at 100ipm is a consistant .0008 - .0009
[18:38:17] <jmkasunich_> do you intend to cut parts at 100 ipm?
[18:38:30] <skunkworks_> No - that was my other question - should I worry about it.
[18:38:46] <skunkworks_> most likely not
[18:38:47] <jmkasunich_> I'd look at the ferror at normal cutting speeds
[18:39:04] <skunkworks_> probably under 20ipm on that machine.
[18:39:56] <jmkasunich_> it is interesting that the error you see matches the distance traveled in one servo period (or half a period)
[18:40:29] <jmkasunich_> if the error is truly just the machine lagging behind the command for a mS, it will have no effect on accuracy at all
[18:41:04] <jmkasunich_> lagging behind by 0.005 isn't gonna make a bad part, because the tool still follows the proper path
[18:41:12] <skunkworks_> I know the error was halfed when I went from 1ms period to .5ms
[18:41:18] <jmkasunich_> being 0.001 off to one side of the path makes a bad part
[18:41:56] <jmkasunich_> you don't care when the metal gets cut, just where ;-)
[18:42:05] <skunkworks_> right
[18:42:44] <skunkworks_> so far the h-bridge is performing nice. no issues but only 10a so far at about 20v.
[18:43:03] <jmkasunich_> heh
[18:43:10] <jmkasunich_> walk before run, right?
[18:43:13] <skunkworks_> right
[18:43:23] <jmkasunich_> when you raise the voltage, you'll need to retune a bit
[18:43:39] <jmkasunich_> doubling the volts is the same as doubling all the gains
[18:43:39] <jmkasunich_> actually, scratch that
[18:43:46] <jmkasunich_> you can change the pwmgen scaling to counter it
[18:43:54] <skunkworks_> yes - a lot - this is only a small servo right not. (at home)
[18:44:01] <skunkworks_> now
[18:44:34] <skunkworks_> I plan on having to retune it from scratch. I could use the experience
[18:45:21] <skunkworks_> I think you saw this http://www.electronicsam.com/images/KandT/servostart/ampmess.JPG
[18:45:43] <jmkasunich_> yeah
[18:46:05] <skunkworks_> other than adding a flywheel - and it is clamped to the table now ;)
[18:46:46] <skunkworks_> that actually changes the tuning :) if the motor can move a bit. duh
[18:51:12] <jmkasunich_> changes the effective inertia
[18:51:25] <jmkasunich_> jmkasunich_ is now known as jmkasunich
[18:54:28] <skunkworks_> how is the digging coming?
[18:55:33] <jmkasunich> its not
[18:55:44] <jmkasunich> today is when I found out that I have a busted pipe
[18:55:54] <jmkasunich> need to get quotes, make arrangements, all that shit
[18:56:09] <jmkasunich> * jmkasunich is having sticker shock
[18:56:10] <skunkworks_> yeck
[18:56:11] <skunkworks_> I bet.
[18:56:24] <skunkworks_> 130 dollars an hour - back hoe rental..
[18:56:29] <jmkasunich> I thought I was just taking a half-day to be here when the county snake guys did their thing (which is free)
[18:57:04] <jmkasunich> instead I paid 300 for a real plumber to snake it and camera it, and I know now I'm looking at thousands to come
[18:57:47] <skunkworks_> so - I guess if he got it cleaned out - you should be good until you can schedule the repairs
[18:57:57] <jmkasunich> yeah
[18:58:11] <jmkasunich> hard to say how fast it will clog up again
[19:05:01] <skunkworks_> depends on how regular you two are... ;)
[19:05:36] <jmkasunich> tmi
[19:14:12] <anonimasu> tmi?
[19:14:56] <skunkworks_> too much information
[19:18:20] <anonimasu> ah