#emc-devel | Logs for 2010-06-11

[00:29:42] <kgartner> I'm wondering if there's a tutorial of some kind on writing new hal components
[00:32:23] <dgarr> http://www.linuxcnc.org/docview/html//hal_comp.html
[00:35:21] <kgartner> thanks, looks like fun :)
[00:47:51] <cradek> what kind of comp are you working on?
[01:21:28] <kgartner> a custom control law to track satelites
[01:22:03] <kgartner> part of research for my masters research
[01:22:12] <Jymmm> Just wear a tin foil hat like everyone else!
[01:23:04] <kgartner> haha, but then I can't watch my direct tv on a boat!
[01:23:37] <Jymmm> your thesis is marine sat tv?! LOL
[01:24:19] <Jymmm> No wonder the kids these days ar Kollage Edjoomakated
[01:24:46] <kgartner> haha, well we sure try to make it as complicated as posible :)
[01:25:33] <kgartner> like anything else it's really more about who's willing to give us money
[01:26:29] <Jymmm> Nobody if you keep watching Sat TV! =)
[01:26:56] <Jymmm> I have yet to see an ad that says "Will pay you to watch The Simpsons"
[01:28:03] <Jymmm> do you plan on compensating for the sway of the water?
[01:28:26] <kgartner> ha, yea, if only. We're trying to convince the company to send us on a big boat so we can profile wave motion... for some reason they think we'll just sit in the sun
[01:28:52] <Jymmm> Well, DUH... they know you too well.
[01:29:07] <Jymmm> I think it was the cooler that said "BEER" that gave it away
[01:29:42] <kgartner> yea, that's pretty much the whole problem, basically it amounts to having an IMU running some kind of kalman filter or similar to figure out roll and pitch and then running the dish motors the opposite ways
[01:30:01] <Jymmm> gyroscope
[01:30:28] <Jymmm> But running a motor constantly for that purpose seems overkill for some reason
[01:30:43] <Jymmm> what about a physical dampner
[01:31:35] <Jymmm> like they use for skyscrapers
[01:32:42] <Jymmm> or Marine Chronometer
[01:32:51] <kgartner> that would probably work if your boat was at the same lat long all the time, but you'd have to adjust it all the time
[01:33:02] <kgartner> the dish needs to be within 3 degrees
[01:33:22] <Jymmm> Sure, but I mean for to compensate for the swells
[01:34:16] <kgartner> like if you're at anchor and just kinda bobbing around?
[01:34:26] <Jymmm> correct
[01:35:15] <kgartner> it would actually be interesting to see if that sort of thing worked... I don't know if that sort of thing has been done before
[01:35:34] <Jymmm> Been done for hundreds of years actually =)
[01:35:43] <Jymmm> http://www.stanleylondon.com/compboxfunct.htm
[01:36:20] <Jymmm> http://en.wikipedia.org/wiki/File:Tuned_mass_damper.gif
[01:36:25] <kgartner> running an azimith/elevation motor is kind of a standard way of doing it, cause then you don't have to worry about setting it for a given lat/long, it's all automatic
[01:36:35] <Jymmm> http://en.wikipedia.org/wiki/Tuned_mass_damper
[01:36:42] <kgartner> haha, yes, just without dish :)
[01:37:42] <Jymmm> I know that yachts already have that capabiity
[01:38:30] <kgartner> oh, like they put tuned dampers in them, to smooth the ride?
[01:39:10] <Jymmm> that I don't know, it was just how I would aproach it. Not leaving it ALL to electronics, but to some mechanical too
[01:41:24] <kgartner> right, so I think the reason the tuned mass dampers work well in bridges and buildings is because those things tend to resonate at certain freq so they can build these devices that basically dampen things over that small range of freq, I'm guessing if you did a FFT of boat vibrations you'd see more than a few resonate peaks
[01:42:47] <kgartner> like I said I'd be interested in seeing if you could get down to less than 3 degrees with really nice bearings and a counter weight
[01:43:27] <Jymmm> Well... GIT ER DONE!!!!
[01:44:07] <kgartner> haha, gotta figure out how to make these componenents first!
[01:45:38] <kgartner> and then once that's done I get to play with my PUMA :)
[01:47:12] <skunkworks> what model puma?
[01:47:28] <kgartner> 500
[01:47:36] <skunkworks> cool
[01:49:18] <skunkworks> there is a new serial kins in emc that needs testing :)
[01:49:43] <kgartner> well, I just happen to have a couple serial kin robot arms laying around
[01:50:19] <kgartner> I was just going to ask how well tested the kinematic and inverse kinematics are
[01:51:29] <skunkworks> lightly tested
[01:52:06] <Jymmm> kgartner: Once you start testing it, we'll know =)
[01:52:22] <kgartner> haha, yea, I guess I've been warned
[01:52:52] <kgartner> I looked through the source and it looked about the same as what I came up with, but I'll have to take a closer look now
[01:56:37] <skunkworks> kgartner: http://juve.ro/blog/puma
[01:56:53] <skunkworks> cradek: that estop rung makes sense now ;)
[02:03:32] <skunkworks> cradek: http://imagebin.ca/img/SqqMqHC.png
[02:03:46] <kgartner> very cool, I have yet to figure out what the difference between the 560 and 500 is
[02:04:28] <skunkworks> the 560 is 6 axis
[02:04:36] <skunkworks> I don't know what the 500 is
[02:05:27] <kgartner> 500 is 6 axis as well
[02:05:48] <kgartner> my manuals for it say '500 series' so they must not be that far apart
[02:06:27] <skunkworks> ah - I think my manual say 500 series also.
[02:06:45] <kgartner> does the 560 have coupling in the wrist axis? ie joint 5 moves joint 6
[02:07:31] <skunkworks> yes - if I understand what you are saying
[02:08:38] <kgartner> yea, so I guess if we have the same manuals they must be pretty much the same
[02:10:01] <skunkworks> this is from the manual http://www.electronicsam.com/images/puma/pumaarm.jpg
[02:12:06] <kgartner> dimensions are in... inches?
[02:12:19] <skunkworks> yes
[02:13:26] <kgartner> okay, yea, those all look the same on mine as well, and I've got the pneumatic stuff
[02:13:51] <skunkworks> pincher?
[02:14:07] <kgartner> I don't have one... but it may have had one at some point
[02:15:03] <kgartner> there's a small (inch or so) offset between the wrist and elbow... if you know what I mean
[02:20:43] <Dave911_> Dave911_ is now known as Dave911
[02:23:10] <skunkworks> is yours made by UNIMATION?
[02:28:30] <kgartner> yea, aren't they all?
[02:29:17] <skunkworks> I thought there was a few companys that made puma arms
[02:29:54] <kgartner> hmm, maybe, who knows, it looks just like the pictures from your site
[05:15:36] <azurra> can I use the Stepconf wizard to create non-trivial kins machines?
[05:15:41] <azurra> Like a Delta robot
[05:16:06] <azurra> XYZ cartesian libertys and 3 joints
[05:16:46] <azurra> i intend to create a a configuration with stepconf and change the kins of that in hal file
[05:17:11] <azurra> i have the .ko file with my new kins, and i know how to change
[05:17:27] <azurra> the problem is that when im creating the machine in step conf
[05:17:42] <azurra> i need to put in axis configuration the Leadscrew pitch rev/mm
[05:18:00] <azurra> and this dont work in my kins
[05:19:50] <azurra> rev/mm is not a linear result
[05:20:03] <azurra> for the axis
[05:20:26] <azurra> any ideas?
[13:40:37] <jepler> aha, is cos(theta) = -cos(theta-180degrees) ?
[13:41:08] <cradek> of course it is
[13:41:33] <cradek> except for maybe zeroes or something
[13:41:34] <cradek> why?
[13:42:21] <jepler> still puzzling over value and dir in andy's comp
[13:42:29] <cradek> ah
[13:43:24] <jepler> it knows the current motor angle and uses 'dir' to make the winding current point either +90 or -90 degrees from the current, then multiplies by 'value'
[13:44:02] <jepler> I was thinking that in normal use you'd have to separate out the sign of the command from the magnitude to get the right phase shift
[13:44:19] <jepler> but as long as the difference is 90 then the sign of value is enough
[13:44:49] <jepler> unfortunately, you can choose a different value than 90 degrees for the phase difference and I think it's wrong for other values
[13:45:14] <jepler> since you'd get e.g., +60 and -120
[15:08:16] <dgarr> for consideration: http://www.panix.com/~dgarrett/stuff/0001-emcrsh-getopt_long-fix-longopts-termination.patch
[15:22:13] <skunkworks> dawned on me why ladder doesn't automatically load in the demo-sim-hal... ( you can load it from the axis menu )
[15:22:38] <skunkworks> duh
[15:30:50] <cradek> yeah it loads and runs automatically, and you can start the gui from the menu
[15:36:43] <CIA-2> EMC: 03jepler 07v2.4_branch * rc341121cd440 10/src/emc/usr_intf/emcrsh.cc: emcrsh: getopt_long() fix longopts termination
[18:01:40] <skunkworks> cradek: I was used to the older sim_cl that automatically started the CL gui
[20:16:01] <alex_joni> we didn't fix http://www.linuxcnc.org/compile_farm/
[20:16:07] <alex_joni> wonder if anyone still uses it
[20:19:26] <alex_joni> ok, fixed
[20:20:23] <alex_joni> but it's probably dead since buildbot
[20:20:28] <alex_joni> jmkasunich: right?
[20:23:06] <jepler> alex_joni: yes I'm pretty sure the compile farm is retired now