#emc-devel | Logs for 2010-10-30

[00:14:25] <skunkKandT> I think I got emc into an odd state. I was running z to 0 and I think it overshot just a little bit and hit the limit switch which sent it into estop, I now have a scrolling 'exceeded negative soft limit on joint 2'
[00:15:38] <cradek> I got that scrolling forever too, just yesterday? it was when I did some weirdo thing with my rotary table (I think I homed it while I had pid unhooked, and then it gleefully flew past the soft limit because it was open loop)
[00:17:03] <cradek> skunkKandT: why not just set the index between steps, so when you home (index only), it always moves by one step?
[00:17:14] <cradek> skunkKandT: you have no choice currently - it WILL unlock during homing
[00:18:03] <cradek> (if you put it in the middle of the slack, you won't know which side of it you're on when you start up)
[00:18:29] <cradek> oh I kept reading - you've got a scheme to deal with that
[00:19:08] <skunkKandT> hmm - that might be an option. it would go up - rotate until it finds the index - rotates back - sets down.
[00:19:26] <cradek> yes
[00:19:37] <cradek> I didn't think of that - you're right - it could just move back
[00:19:47] <cradek> that's sure simpler?
[00:19:48] <skunkKandT> right - the offset
[00:20:03] <skunkKandT> I think so. acomplishes the same thing
[00:20:09] <skunkKandT> much easier
[00:20:13] <cradek> then if you g0 g53 a0 before you shut down, you'll never have to mess with it
[00:20:26] <cradek> (or g28 even)
[00:20:45] <andypugh> I feel such a fool.
[00:21:00] <cradek> uh-oh
[00:21:03] <andypugh> I have been wondering why my bldc motor won't commute.
[00:21:29] <andypugh> Turns out that I can't decide how many degrees are in a full circle.
[00:21:43] <cradek> pi! 2pi! 400!
[00:21:58] <cradek> 359!
[00:22:06] <andypugh> I have 1024 in a resolver circle, 360 in the bldc-sine component and 1 in the 8i20 driver.
[00:22:07] <cradek> sqrt(2)?
[00:22:20] <cradek> haha
[00:22:25] <andypugh> It was that last bit that was causing the problem
[00:22:53] <andypugh> The bit-dither in the resolver output was casing utterly random movement n the 8i20 driver
[00:23:23] <andypugh> Now the question is, what is the _right_ number.
[00:24:21] <andypugh> It's -1 to 1 for current, should I have 0 to 1, -1 to 1, 0 to 360, -180 to 180 for the phase-angle input?
[00:25:23] <cradek> I have no idea :-)
[00:26:21] <cradek> spindle encoder scaling is 1.0 increase per rotation - nothing else really deals with rotations
[00:26:46] <cradek> might want to consider how the future canonical hal resolver interface will look, and match that
[00:27:27] <andypugh> _Is_ there a future canonical HAL resolver interface?
[00:27:52] <cradek> heh, I hope there is, since the 7i49 or whatever it's called is coming
[00:28:21] <andypugh> I wonder who is writing the driver for that?
[00:28:32] <cradek> ummm, I volunteer ... anyone else
[00:28:40] <cradek> probably/hopefully seb?
[00:29:46] <cradek> if I ruled the universe (I don't) I think I'd want it to be just like an encoder with index at "0" with an additional hal pin that shows the absolute position scaled 0..1
[00:30:15] <andypugh> It sounds like an easy one, so perhaps even me. The 8i20 has not been easy, having a second layer of abstraction and numerous buggy drivers. (to the extent of several code tear-ups, wondering how I broke it, and then finding it was broken in the firmware. Grumble)
[00:30:54] <cradek> wish you could come to fest...
[00:31:07] <andypugh> Yes, 0-1 makes sense, saves a lot of dividing.
[00:31:50] <andypugh> I will do it that way, and make bldc-sine talk in terms of fractional turns too.
[00:33:43] <andypugh> In theory I could come to the fest, but it is a lot of money.
[00:33:55] <cradek> I think 0-1 is the scheme you can choose where nobody will say it's clearly crazy :-)
[00:34:01] <andypugh> Actually, it is less money than last time I looked. £383 now.
[00:35:04] <cradek> wow, that seems not bad at all
[00:35:13] <cradek> although, your money is worth more than mine :-)
[00:35:23] <cradek> does that get you to wichita?
[00:35:56] <andypugh> Yes, though not necesarily the correct Wichita
[00:36:10] <cradek> wichita KS, lower middle of the US
[00:36:19] <andypugh> (I guessed at Wichita, KS, not Wichita Falls.
[00:36:28] <cradek> yep that's the one
[00:37:24] <andypugh> Seen pcnx's picture?
[00:40:45] <andypugh> Not what I thought he was on about.
[00:40:48] <andypugh> Sorry.
[00:40:59] <cradek> heh
[00:41:05] <cradek> "guess what I think is wrong!!"
[00:41:15] <cradek> "nope!! guess again!!"
[00:42:17] <andypugh> Yeah. I do think I have had the "can't do that without being homed, when I am. But I think it comes down to managing to unhome a joint in a machine with more joints than axes
[00:42:44] <cradek> yeah, wouldn't surprise me if that was screwy in master
[00:44:41] <skunkKandT> hmm - z is a little tighter on one end of the travel vs the other.
[00:44:46] <skunkKandT> yecky
[00:44:56] <cradek> eww
[00:45:41] <skunkKandT> you can hear it in the helix program on youtube. when z shuttles home - it sounds different initally
[00:46:06] <skunkKandT> it sounds like it is going faster - but in reality it is just the servo working harder.
[20:47:33] <jepler> logger_dev: I liked you better when you'd just disappear after pinging out
[20:47:33] <jepler> I'm logging. I don't understand 'I liked you better when you'd just disappear after pinging out', jepler. Try /msg logger_dev help
[20:49:09] <SWPadnos> logger_dev, don't take things so literally, geez!
[20:49:09] <SWPadnos> I'm logging. I don't understand 'don't take things so literally, geez!', SWPadnos. Try /msg logger_dev help
[21:14:58] <skunkworks> heh