#emc-devel | Logs for 2007-07-05

[13:33:52] <jepler> $ halcmd show pin axis.*.motor-pos-cmd
[13:33:52] <jepler> Component Pins:
[13:33:52] <jepler> Owner Type Dir Value Name
[13:33:52] <jepler> 32770 float OUT 0 axis.0.motor-pos-cmd ==> Xpos
[13:33:52] <jepler> 32770 float OUT 0 axis.2.motor-pos-cmd ==> Zpos
[13:35:15] <cradek> cool
[13:35:45] <jepler> something's funky
[13:35:51] <jepler> the GUI is unresponsive
[13:35:51] <jepler> I wonder why
[13:36:28] <jepler> no, it's more than that -- it doesn't run at all
[13:36:29] <jepler> :-P
[13:36:44] <cradek> -cool
[14:10:14] <jepler> I wonder why
[14:10:45] <cradek> once I got a "can't initialize motion" problem that acted that way
[14:11:11] <cradek> check dmesg carefully
[14:13:01] <jepler> it's sim so I see "dmesg" on the terminal
[14:46:59] <jepler> skunkworks: next time you're at your machine with pluto, you should test the watchdog
[14:47:30] <skunkworks> Sure. will man tell me how?
[14:47:42] <jepler> skunkworks: get halcmd running in the terminal. then do something that will set the motor spinning. then in halcmd type: "stop"
[14:47:59] <skunkworks> ok
[14:48:04] <jepler> this will make the PC stop communicating with the pluto board, and about 6.5ms later the pluto should stop sending its PWM output signal
[14:48:29] <skunkworks> I think I can handle that ;)
[14:48:48] <skunkworks> so the watchdog is transparent? no pins need to be hooked up?
[14:48:52] <jepler> yep
[14:48:54] <skunkworks> Nice
[14:49:27] <jepler> you can disable it with a 'loadrt' parameter -- for instance, if you wanted to run a 10ms servo period you'd have to disable it
[14:50:01] <skunkworks> ah - makes sense.
[14:51:17] <jepler> but otherwise it gets enabled as soon as the PC starts writing data (pluto-servo.write function in HAL) to the pluto, and activates after 6.5ms without hearing from the PC
[14:51:25] <anonimasu> hm
[14:51:38] <anonimasu> skunkworks: what wattage does your boards handle?
[14:51:41] <anonimasu> err the ones you have..
[14:51:49] <anonimasu> I need 900w.. :) peak
[14:54:51] <skunkworks> that should not be a problem. It should do into the 150v 20a
[14:55:58] <skunkworks> continuous - with a fan ;)
[14:56:01] <skunkworks> big fan
[14:59:05] <anonimasu> nice
[14:59:12] <anonimasu> can you route thoose boards on a mill?
[15:00:05] <anonimasu> * anonimasu is very curious about a plutoP and 3 of your drives
[15:00:19] <anonimasu> building 3 of them
[15:00:35] <skunkworks> Yes - I have been milling them on our gantry.
[15:00:47] <skunkworks> Remember - no current limiting.
[15:00:56] <anonimasu> oh :s
[15:01:33] <anonimasu> skunkworks: will you have that later?
[15:02:05] <anonimasu> that's a big thing for me ;)
[15:02:12] <skunkworks> Maybe...
[15:02:28] <skunkworks> it gets messy - so I thought I would try this first.
[15:02:47] <anonimasu> hm, I guess you could have a opamp and a shunt.. on the input wire..
[15:03:09] <anonimasu> and run feedback to the pluto...
[15:03:24] <anonimasu> or just kill it off.. when current gets too high
[15:03:33] <anonimasu> and send a signal to the pluto..
[15:03:52] <anonimasu> though if you have feedback you can soft limit it..
[15:04:27] <anonimasu> sorry if im ranting
[15:07:23] <anonimasu> would there be enough space on the pluto-p for a adc?
[15:07:57] <anonimasu> because then you could run current feedback..
[15:12:31] <jepler> anonimasu: nothing of the kind is planned
[15:13:04] <jepler> it doesn't have any internal ADC so you're stuck building something external
[15:13:23] <jepler> and there is not a lot of capacity left inside the pluto to add additional logic
[15:16:53] <jepler> to use current feedback filter the PWM output to analog and hook it up to some IC with current control as the reference voltage input (I think, but I'm not 100% sure, that this is how the mesa servo amps work)
[15:20:16] <anonimasu> jepler: that was the idea..
[15:21:02] <anonimasu> jepler: maybe a mesa would be better..
[15:21:49] <anonimasu> jepler: though I guess a slow fuse would work..
[15:22:20] <anonimasu> jepler: how do commercial servodrives handle it?
[15:22:21] <anonimasu> fuses?
[15:25:54] <jepler> I don't know
[15:25:59] <jepler> * jepler <-- strictly bush league
[15:26:30] <anonimasu> bush league?
[15:26:44] <jepler> amateur
[15:26:45] <jepler> hobbyist
[15:26:48] <anonimasu> ah..
[15:26:50] <anonimasu> well I guess a fuse would do..
[15:27:02] <anonimasu> I doubt I could blow a fuse at the big mill at work.
[15:28:16] <anonimasu> in general isnt ut true that if your motors draw more current then the peak rating + a bit your are fscked..
[15:28:18] <anonimasu> ;)
[15:28:31] <anonimasu> or your machine is badly designed..
[15:28:39] <anonimasu> *sorry this is the dev chan :/*
[15:29:59] <skunkworks> My servos have peak rating of 88Amps
[15:31:40] <anonimasu> *grin*
[15:32:07] <jepler> anonimasu: yes, the #emc channel would be a better place for this discussion
[18:05:23] <jepler> cradek: if the gantrykins thing you were mentioning works, that is a better solution than what I was trying to do
[18:05:47] <jepler> I'm still not sure why it's not working..
[18:07:37] <cradek> you think the problem is the nonsequential joint numbers in hal?
[18:11:56] <jepler> it seems like the problem is in task, not motion
[18:12:10] <jepler> I can jog just fine, I just can't switch to MDI or AUTO
[18:12:31] <jepler> I didn't look into it after I decided it probably wasn't motion
[20:56:24] <skunkworks> barry had to go back in - the absess was growing agian.
[21:03:47] <jepler> tell him not to make a habit of it, will you?
[21:04:37] <skunkworks> I will try.. He doesn't listen very well..
[21:05:00] <jepler> no, cats tend not to
[21:05:04] <jepler> bbl
[21:05:04] <skunkworks> :)
[21:18:18] <SWPadnos> cradek, how hard do you think it would be to adda script (and possibly a desktop icon) to the live CD to create the necessary /dev/rt* files and run the latency test?
[23:35:13] <jepler> http://pastebin.ca/605127
[23:36:04] <jepler> skunkworks: http://pastebin.ca/605127
[23:36:59] <jepler> not that you care about step generation, I suppose
[23:37:44] <skunkworks> holy crap.
[23:37:46] <skunkworks> nice
[23:38:46] <jepler> I don't know how good its step waveform actually is yet
[23:38:57] <skunkworks> well - we do have the gantry with steppers...
[23:39:38] <jepler> oh really
[23:39:44] <skunkworks> really :)
[23:40:02] <jepler> does it need high step rates?
[23:40:41] <skunkworks> only z - it has an input scale of 10000
[23:40:49] <skunkworks> but it would be nice to play with.
[23:41:00] <jepler> takes step+direction input?
[23:41:00] <skunkworks> great - something else to play with..
[23:41:03] <skunkworks> yes
[23:41:44] <skunkworks> like I have time for that ;)
[23:42:03] <jepler> indeed
[23:42:37] <cradek> SWPadnos: the easiest way for me would be if you would make a package that installs your program (script) and also a menu item to run it
[23:45:40] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/simulated-step-waveform.png -- showing movement at 2 different velocities (left) then a direction change with space before and after (right). green is step, red is direction.
[23:46:15] <jepler> I don't actually have anything that takes step+direction inputs or I might actually test it myself
[23:46:38] <jepler> bbl
[23:59:21] <jepler> I wonder why the feedback velocity is 5% off from the requested velocity
[23:59:44] <jmkasunich> this is pluto stepgen?
[23:59:53] <cradek> hi guys
[23:59:56] <jmkasunich> hi