#emc-devel | Logs for 2008-03-05

[02:55:07] <Guest713> Guest713 is now known as skunkworks_
[09:53:55] <christel> [Global Notice] Hi all! We experienced some rather heavy ddos attacks earlier this morning causing quite a lot of disturbance to the network, the problems should be under wraps now (thanks to the OSL and NERO) and we apologise for any inconvenience caused. Hope you have a nice day and thank you for using freenode!
[12:38:49] <cradek_> cradek_ is now known as cradek
[15:57:46] <fenn_> fenn_ is now known as fenn
[16:09:58] <CIA-22> EMC: 03cradek 07TRUNK * 10emc2/tests/ccomp/lathe-comp/expected: w tool lengths changed this output. new output is correct.
[19:25:39] <CIA-22> EMC: 03lerman 07TRUNK * 10emc2/src/emc/rs274ngc/interp_read.cc: fix bug: [1709015] comments in O-call lines disrupt parameters
[19:55:20] <CIA-22> EMC: 03lerman 07TRUNK * 10emc2/src/emc/rs274ngc/interp_convert.cc: fix bug:[1807740] PRINT command does not evaluate named parameters correctly.
[20:40:50] <SWPadnos> hmmm. cradek, I just had a thought, and you can test it since you have a stepper machine
[20:41:13] <SWPadnos> of course, I could be forgetting some old conversation where this idea was rejected
[20:41:37] <SWPadnos> hmmm. damn. it's not as easy as I want :(
[20:41:57] <SWPadnos> the brainstorm was that stepgen shouldn't be necessary in EMC
[20:42:41] <SWPadnos> freqgen should be able to do the job, because motion already does accel profiles
[20:45:55] <skunkworks> you're talking hooey again
[20:46:10] <SWPadnos> hooo-eeeee!
[21:07:28] <alex_joni> SWPadnos: but freqgen only gives you vel feedback
[21:07:33] <alex_joni> not position feedback
[21:07:49] <alex_joni> so if you want to do it this way, you need a PID, and a counter on the steps
[21:07:51] <SWPadnos> yeah, that's one of the places where you get more HAL complexity
[21:08:03] <alex_joni> and there's no gain for users
[21:08:12] <SWPadnos> ddt on motion output, counter on feedback should do it
[21:08:22] <alex_joni> except that you're back at square 1, where they need to tune stepper systems
[21:08:25] <SWPadnos> the gain is that the confusion surrounding STEPGEN_MAX* settings goes away
[21:08:36] <alex_joni> which is something we gladly stepped away from emc1
[21:08:36] <SWPadnos> small gain :)
[21:08:47] <alex_joni> huge backstep
[21:09:40] <cradek> dude, talk to jmk, and leave my stepper machine out of it
[21:09:47] <SWPadnos> heh
[21:09:59] <alex_joni> I'd advice against that
[21:10:12] <alex_joni> he'll get the biggest trout in a 500 miles radius
[21:10:17] <alex_joni> and slap you silly with it
[21:10:28] <SWPadnos> luckily he's 700+ miles from me :)
[21:10:33] <cradek> I'm sure he'd love to hear about how everyone should change around what works so well right now
[21:10:46] <alex_joni> cradek: he already plans to do that
[21:10:51] <alex_joni> but that's a different topic :)
[21:10:56] <cradek> you mean the trout thing?
[21:11:03] <SWPadnos> well, this was one of those thoughts that comes to you when you're preparing breakfast (at 3:00 in the afternon)
[21:11:03] <cradek> * cradek is so lost
[21:11:08] <alex_joni> no, making all users change configs
[21:11:09] <SWPadnos> changing everything :)
[21:11:25] <cradek> ah, nothing wrong with that, if it's for a good reason
[21:11:28] <alex_joni> cradek: changing axis.* pins from motion with joint.*
[21:11:34] <cradek> like to make max work better
[21:11:35] <cradek> :-)
[21:11:38] <alex_joni> pin names..
[21:20:13] <jepler> SWPadnos: you can run stepgen without setting maxvel or maxaccel; they default to "infinite". but skunkworks tested and found that specifying a finite accel "worked better"; I don't know why or remember the specifics of what seemed better to him.
[21:20:13] <alex_joni> I think infinite vels especially together with backlash moves will result in bad things for the motor
[21:20:13] <SWPadnos> the real problem there is when the MAX_* for the TP are actually at the same value as the stepgen effective max (which could be limited due to BASE_PERIOD)
[21:20:13] <alex_joni> you won't get a warning then..
[21:22:03] <skunkworks> with an finite accel - My machine would not stall.. With an infinite accel - my machine would constantly stall during rapids.
[21:22:26] <skunkworks> (running it at the edge of usefulness)
[21:23:23] <skunkworks> (pushing the envelope)
[21:26:30] <skunkworks> I think SWPadnos had tried to explain to me why this was probably happening - but I don't remember why
[21:27:28] <SWPadnos> me either :)
[21:27:39] <SWPadnos> actually, I may remember
[21:28:22] <SWPadnos> there was a difference between zero accel, "finite" accel, and really huge accel (effectively infinite)
[21:28:51] <SWPadnos> I think the deal was that with 0, there is no accel applied - stepgen just goes to the new frequency immediately
[21:29:58] <SWPadnos> if the accel limit is high enough that stepgen can get there in one servo cycle, then the make_pulses routine will actually smoothly ramp to the new frequency (within a single servo period)
[22:36:00] <alex_joni> good night all
[22:36:19] <skunkworks> night alex