#emc-devel | Logs for 2010-10-28

[00:24:37] <skunkworks> cradek: http://www.electronicsam.com/images/KandT/conversion/curvicfoot.jpg
[00:24:56] <skunkworks> http://www.electronicsam.com/images/KandT/conversion/curvictogether.jpg
[00:30:06] <cradek> heh, the scale on that is bigger than I expected...
[00:30:28] <cradek> so the sitting down on this locking ring is what actually positions it accurately?
[00:30:48] <cradek> I wonder if you want some lash in the gears on purpose for that to work well
[00:37:53] <jepler> skunkworks: what is *that*?
[00:39:26] <skunkworks> that is what indexes the B axis - 72 teeth 5 degree index
[00:40:01] <skunkworks> yes - that is what positions the axis accurately
[00:40:33] <jepler> I see
[00:41:18] <skunkworks> cradek: good point. I suppose we could disable the servo amp/pid after it gets in position. didn't really think about that/.
[00:43:40] <skunkworks> what would you have to disable so that when it gets reactivated it doesn't try to jump.
[00:47:21] <skunkworks> unless we got the backlash perfectly inside its slack
[00:47:33] <skunkworks> gear train slack
[00:57:44] <skunkworks> maybe it isn't a problem. If we home to an index and figure out the offset so that it is right in the middle of the gear train slack.
[01:19:14] <KimK> skunkworks: That can be dad's next project, to design and build a gearbox suitable for continuous backlash-free rotation, and get rid of the curvic coupling. "Your mission, should you decide to accept it...", lol. No, maybe not, that is a pretty big platter you've got there, and you don't want to have to baby it while cutting. Tempting, but nevermind.
[01:21:32] <KimK> skunkworks: BTW, curvictogether.jpg is a nice pic, as it shows the necessary gaps on the tops and bottoms, it should only seat on the tapers, and it does.
[01:30:41] <skunkworks> KimK: that would be nice - but I don't know if that is needed. ;)
[03:21:50] <cradek> yes once you get it right, it should stay right - but seems finicky
[13:53:57] <skunkworks> cradek: should be an interesting experiment.
[13:54:17] <skunkworks> how do other locking rotorys get handled. I could see a wind-up issue
[13:54:36] <skunkworks> (if it is a little off)
[14:04:50] <cradek> I agree - I would not use I gain. not sure how servos on locking axes should work.
[14:06:08] <cradek> on my locking indexer it a quarter turn of the worm backlash, and the indexing lock is what did the final orientation. so it would be fine, even with rough homing.
[14:06:55] <cradek> I think yours would need slack to work right, too.
[14:08:10] <cradek> you could disable the amp when the table is down, but I agree you might get jumps when reenabling it.
[14:18:54] <skunkworks> I don't think it is a 1/4 turn - but there is some backlash in the gear train
[14:20:20] <cradek> I bet that's good
[14:20:25] <cradek> don't fix it this time :-)
[14:21:14] <skunkworks> that was never designed to be 0 backlash - no adjustments anywhere.
[14:21:17] <skunkworks> :)
[14:23:20] <skunkworks> we could add backlash... it is going to be run with a chain :)
[14:33:18] <skunkworks> The only other thing I need to worry about is if there is an even number of rotations of the input shaft for one complete rotation of the table.
[14:33:44] <cradek> why?
[14:33:58] <cradek> oh you mean integral number of rotations - for index
[14:34:04] <skunkworks> right
[14:34:35] <skunkworks> they had a resolver in the original gearbox - but that was driven by some extra gears. So I am not sure.
[14:35:48] <cradek> yeah, hope it is integral...
[14:35:53] <cradek> probably it is?
[14:35:58] <skunkworks> heh - no clue
[15:54:07] <skunkworks> looks like the resolver was geared 2:1
[15:54:37] <skunkworks> so that is promising