#emc-devel | Logs for 2007-04-29

[00:07:57] <mschuhmacher> cradek: the System is runnig with the new Kernel
[00:08:12] <skunkworks> 4 processors?
[00:08:32] <mschuhmacher> all 4 are shown
[00:08:34] <mschuhmacher> :)
[00:09:00] <skunkworks> Nice
[00:09:11] <skunkworks> now time for the latancy test.
[00:10:14] <mschuhmacher> first I want to recompile the rtai
[00:11:09] <mschuhmacher> with the current kernel-settings
[00:23:22] <mschuhmacher> the latency test is running ovl max 41048
[00:23:41] <cradek> ouch, that's fairly bad
[00:24:59] <mschuhmacher> I have to recompile the kernel
[00:25:26] <mschuhmacher> I think I over-optimized it maybe
[00:25:57] <mschuhmacher> but I want to run servos
[00:27:37] <cradek> you're using a 4 processor machine to run servos? that seems like a terrible waste of a machine
[00:27:42] <skunkworks> I really don't think you need a 4 proccessor computer to run servos ;)
[00:27:48] <cradek> yeah like he says :-)
[00:27:55] <skunkworks> ;)
[00:28:17] <mschuhmacher> I buyed it cheap
[00:28:23] <cradek> what cpus?
[00:28:23] <jmk-solo> maybe he's running a whole plant full of servo machines
[00:28:34] <jmk-solo> one mill per CPU
[00:28:36] <mschuhmacher> 4x Xeon 2.4GHz
[00:28:42] <jmk-solo> wow
[00:29:04] <mschuhmacher> http://www.filpro.de/Photo/Screenshot.png
[00:29:45] <mschuhmacher> the maschine is fast, but slow graphics and no soundcard
[00:29:54] <jmk-solo> a server in other words ;-)
[00:30:03] <cradek> yeah
[00:30:22] <mschuhmacher> redundant power-suply an so
[00:30:34] <jmk-solo> bet its loud ;-)
[00:31:06] <mschuhmacher> I promised my neighbour to shutdown at ten
[00:31:11] <jmk-solo> lol
[00:31:18] <mschuhmacher> :) a joke
[00:31:56] <mschuhmacher> its not so loud as the vacuum-cleaner
[00:33:08] <mschuhmacher> but the last two days I used earplugs
[00:47:35] <skunkworks> if you guys need to incorperate any of this into emc - I don't have an issue ;) http://www.electronicsam.com/images/KandT/xyz.txt
[00:48:21] <cradek> IF ABS(xcomp - xstep) >= .0039 THEN
[00:48:25] <cradek> haha
[00:48:32] <cradek> ummm I don't see a license statement, so we can't use it
[00:49:07] <skunkworks> it was pretty rough ;)
[00:49:25] <cradek> what is it?
[00:50:27] <skunkworks> it ran steppers out of the printer port - did g0,1,2,3
[00:50:35] <skunkworks> phase iirc
[00:50:48] <cradek> ah, cool
[00:50:49] <cradek> arcs even? wow
[00:51:31] <skunkworks> I don't remember if this was where I stopped or if I had something newer
[00:53:20] <skunkworks> I remember after seeing it all but stopping at end points - I needed some sort of read-ahead/blending.. Then lost interest as I couldn't wrap my head around it
[00:54:34] <cradek> did you use no acceleration at all? that must have limited your speed (I guess you didn't have much choice without realtime)
[00:55:26] <skunkworks> I had used loops for accelleration (counting down) but I don't really see it in there. So I assume I had done more developing than that.
[00:55:45] <cradek> ah
[00:56:15] <skunkworks> FOR accdelay = 1 TO 0
[00:56:22] <skunkworks> started it ;)
[00:56:39] <skunkworks> daisy?
[00:57:01] <cradek> yes daisy
[00:57:34] <skunkworks> :) ok.. what is daisy?
[00:58:15] <cradek> it's pretty obvious once you look at it
[01:06:14] <jmkasunich> it is?
[01:06:26] <jmkasunich> doesn't seem that obvious to me
[01:07:35] <jmkasunich> it also runs very strangely
[01:08:03] <skunkworks> crap - the server motherboard doesn't have a speaker. (wanted to run sim inch)
[01:08:13] <skunkworks> I had to look at it twice.
[01:09:01] <skunkworks> jmkasunich: you have to look down farther in the program :)
[01:09:14] <jmkasunich> I see the comments, it plays music thru the steppers?
[01:09:39] <skunkworks> I would assume you could hear it on siminch also. assume any ways.
[01:09:41] <jmkasunich> but it runs very strangely - slow for the first move or two, then finishes the rest in about a half a second
[01:11:02] <jmkasunich> on sim-inch, I got a following error after two notes
[01:11:13] <cradek> it runs for about 30 secs
[01:11:28] <cradek> hm, I ran it on trunk...
[01:11:32] <jmkasunich> not here it doesn't
[01:12:27] <cradek> ah it requires 4 axes
[01:12:40] <cradek> wrong things happen if you move undefined axes
[01:12:43] <cradek> that might be a bug
[01:12:46] <skunkworks> ah
[01:14:58] <cradek> you'll also have to set the scales on line 81
[01:15:45] <cradek> (they're set for max)
[01:47:34] <mschuhmacher_> good night all
[02:14:22] <petev> jmkasunich, I back and did the parallel field torque test
[02:14:26] <petev> it worked very well
[02:14:39] <petev> a little more/less and rotation one way or the other
[02:15:16] <petev> turns out 330 degree offset which is 180 more than I calculated from the glentech docs
[02:15:24] <petev> so somebody got something screwed up
[02:15:43] <jmkasunich> could be active high vs. active low halls or something
[02:16:02] <petev> the halls have +/-, but the drive is single ended
[02:16:16] <petev> the glentek doc referenced the +, and that's the one I used
[02:16:29] <petev> but they had so many other mistakes, who knows
[02:16:36] <jmkasunich> to be honest, I think you are doing very well getting non-AB motors to run on AB drives - they don't go out of their way to make it easy, they want you to buy AB everything
[02:16:41] <petev> the 180 is a bit of comfort I guess
[02:17:06] <petev> if it would have been some other wierd number I would really be questioning it
[02:17:18] <petev> the motor is running very nicely now
[02:17:21] <jmkasunich> cool
[02:17:23] <petev> at least as good as the AB motor
[02:17:30] <petev> now to make some more cables
[02:17:40] <petev> this was a learning experience for sure
[02:17:50] <jmkasunich> I learned a bit too
[02:18:00] <petev> I have never found an AB cable that didn
[02:18:03] <jmkasunich> I just hope I remember it when I start my brushless project
[02:18:05] <petev> match their docs
[02:18:22] <petev> I think I learned more than you did on this one
[02:18:28] <petev> you were pretty much helping me
[02:18:40] <petev> I'm sure I'll remember after wasting so much time
[02:18:58] <petev> just give a yell when you're doing yours
[02:19:03] <jmkasunich> ;-)
[02:19:04] <petev> I ave a bunch of notes as well
[02:19:05] <jmkasunich> will do
[02:41:23] <a-l-p-h-a> jmkasunich, sup?
[02:41:38] <a-l-p-h-a> how's the mesa stuff?
[02:42:34] <jmkasunich> it would be faster if I could concentrate, I'm easily distracted
[04:06:40] <petev_> jmkasunich, u still up?
[04:07:23] <jmkasunich> who me?
[04:07:44] <petev_> I was just drawing some diagrams while taking a break to watch the game
[04:08:09] <petev_> and if I didn't make a mistake, it seems like offesting the hall angle 180 deg makes the motor run the other direction
[04:08:26] <petev_> so why does AB make you jump throught hoops to match the encoder?
[04:08:37] <petev_> seems like you could just offset the hall
[04:08:45] <jmkasunich> no clue
[04:09:03] <petev_> does that sound right about the 180 reversing the direction?
[04:09:09] <jmkasunich> yeah
[04:09:40] <petev_> so I think the only issue to begin with was the cable foo and the hall not matching the windings
[04:10:03] <jmkasunich> waitaminnit
[04:10:21] <jmkasunich> offsetting by 180 degrees is not the same as having the direction backwards
[04:10:37] <jmkasunich> the directions of all three, winding, hall, and encoder, need to match
[04:10:49] <petev_> agreed
[04:10:54] <jmkasunich> only then can you reverse it by changing the offset by 180 degrees
[04:10:57] <petev_> but assume encoder is swapped
[04:11:09] <petev_> if you offset hall by 180, wouldn't that fix it?
[04:11:14] <jmkasunich> if something is going the wrong direction, no amount of offset will fix it
[04:11:54] <petev_> oh, I think I'm not taking into account the rotation of the field
[04:12:04] <petev_> the intial move you be the opposite direction
[04:12:16] <petev_> but they it would still follow the field the wrong way
[04:12:19] <petev_> they^then
[04:12:58] <petev_> you^would
[04:13:00] <petev_> man I can't type
[12:13:02] <lerman_> lerman_ is now known as lerman
[15:24:11] <cradek> now I'm interested in this tool length/probe question
[16:25:51] <jmkasunich> sounds dangerous
[16:28:20] <cradek> ?
[16:30:29] <jmkasunich> getting interested in something ;-)
[16:31:23] <jmkasunich> I need to mow the grass, and I'm not interested in that at all
[16:31:45] <cradek> here too - haven't mowed this year yet
[16:32:01] <jmkasunich> same here
[16:32:07] <jmkasunich> getting pretty ragged looking
[16:32:12] <cradek> and there's been a LOT of rain :-)
[16:32:20] <jmkasunich> its been very wet here too
[16:32:28] <cradek> I like it when it looks long, green, and healthy like that
[16:32:38] <cradek> but it gets out of control after a while
[16:32:51] <jmkasunich> we have lots of weeds
[16:33:07] <jmkasunich> they're green, so when you mow them, they don't stand out so much ;-)
[16:33:07] <cradek> I think the grass is so long even the dandelions can't grow here
[16:33:28] <cradek> yeah, if you cut it all the same height, it's grass!
[16:33:32] <jmkasunich> right
[16:36:14] <jmkasunich> I knew this existed somewhere: http://issaris.org/~takis/projects/rtai/livecd/
[16:36:27] <jmkasunich> 8 meg iso, bootable CD with latency test
[16:36:43] <jmkasunich> take it to the local computer store before you buy....
[16:37:11] <cradek> yeah, but if it's not even in graphics mode and does no disk io, it's minimally useful
[16:37:20] <cradek> I'd take the emc cd instead
[16:37:33] <cradek> (for our application)
[16:37:43] <jmkasunich> true
[16:38:03] <cradek> off to breakfast
[16:38:05] <jmkasunich> however, most computer stores wouldn't let you use either one
[16:38:15] <jmkasunich> afraid their silly MS systems might get infested
[16:38:43] <cradek> hey that's very important if you don't have install media and you have to buy a new computer if it gets messed up
[16:39:06] <cradek> (^^ only half a troll)
[16:39:09] <cradek> bbl
[16:51:49] <tomp> that rtai iso is only the kernel & latency tests ( thats its purpose, not for dev or using ) & 'might get infested'//'might get cured'
[17:14:18] <petev> jmkasunich, I figured out where the 180 degree hall offset came from. The AB chart was back EMF and the Glentek chart was the winding voltages.
[17:18:02] <lerman_> lerman_ is now known as lerman
[17:44:16] <lerman_> lerman_ is now known as lerman
[18:10:05] <lerman_> lerman_ is now known as lerman
[18:32:18] <lerman_> lerman_ is now known as lerman
[19:03:05] <lerman_> lerman_ is now known as lerman