#emc-devel | Logs for 2010-10-31

[00:01:04] <ries_> ries_ is now known as ries
[20:44:38] <CIA-31> EMC: 03cmorley 07v2.4_branch * r57ac6f7acdf0 10/src/emc/usr_intf/pncconf/pncconf.py: clean up generated HAL file text for jogging signals
[20:44:39] <CIA-31> EMC: 03cmorley 07v2.4_branch * r7098f4f04396 10/src/emc/usr_intf/pncconf/pncconf.py: change touchy's quillup signalname
[20:44:40] <CIA-31> EMC: 03cmorley 07v2.4_branch * r65371b9d28e3 10/src/emc/usr_intf/pncconf/pncconf.py: use constants for descibing GUIs. fix Touchy signal names
[20:44:41] <CIA-31> EMC: 03cmorley 07v2.4_branch * rb8fecffb8c1c 10/src/emc/usr_intf/pncconf/pncconf-help/ (help-extcontrols.txt help-gui.txt): Add info about TOUCHY
[20:44:41] <CIA-31> EMC: 03cmorley 07v2.4_branch * r16a80ca82e19 10/src/emc/usr_intf/pncconf/pncconf.py: fix error in hal file with default selected-jog-incr
[20:44:45] <CIA-31> EMC: 03cmorley 07v2.4_branch * r00d512d30841 10/src/emc/usr_intf/pncconf/ (pncconf.glade pncconf.py): Add Touchy GUI choice. Fix spindle-at-speed code
[21:56:12] <ries_> ries_ is now known as ries
[22:22:32] <jepler> hm, the documentation apparently says that ABC axis motion is accepted during canned cycles but it looks like they're pretty emphatically not
[22:22:44] <jepler> well, accepted but ignored :/
[22:23:42] <andypugh> From the sample file, it seems like they are ignored until you exit canned cycle mode?
[22:31:50] <jepler> could be
[22:32:05] <jepler> seems to me that if they're not going to be used by the canned cycle, the line should be rejected altogether
[22:32:29] <jepler> not just make a spooky motion to a different C coordinate after changing to some other motion type
[22:33:07] <andypugh> Sorry, I was unclear. I have noidea if they are then acioned after the canned cycle.
[22:34:21] <andypugh> What I meant was, you migh (just) be able to accept that G98 G81 X0 Y0 C10 R2 F20 (say) might ignore the C10, but the sample code had the C move on a separate line.
[22:35:29] <andypugh> I would expect that if C moves are not valid in a Canned Cycle, then they should at least have a sticky G0 or G1.
[22:57:11] <cradek> ick
[22:57:29] <cradek> wonder what he expects that C word to do!
[23:07:33] <cradek> I have a dim memory of an ngc doc that said ABC are accepted on canned cycles, but they can only respecify the existing position (cause no motion)
[23:11:43] <skunkKandT> hey
[23:11:55] <JT-Hardinge> hey
[23:11:58] <cradek> hey
[23:12:13] <skunkKandT> ok - now that I have your attention ;)
[23:12:26] <skunkKandT> I have come across an oddity
[23:12:39] <cradek> you mean JT?
[23:12:52] <cradek> (he's not that odd)
[23:13:02] <JT-Hardinge> he must mean me
[23:13:19] <skunkKandT> if I set the counter mode of the mesa encoder counter to X4 mode - it seems to loose counts.
[23:13:23] <cradek> yeah, couldn't be me
[23:13:31] <skunkKandT> sorry - not that fast ;)
[23:13:50] <cradek> but x4 is the default
[23:13:55] <skunkKandT> so - if I turn my mpg one turn - it might be off by 2
[23:13:58] <skunkKandT> or more
[23:14:10] <skunkKandT> sorry - not X4.
[23:14:41] <skunkKandT> but if I have it in X4 (;)) it counts perfectly each turn
[23:15:36] <cradek> weirdly I'm not seeing an option for x4
[23:15:43] <skunkKandT> oh
[23:15:57] <skunkKandT> setting hm2_[HOSTMOT2](BOARD).1.encoder.01.counter-mode
[23:16:03] <skunkKandT> to 1 seems to do it.
[23:16:25] <cradek> (bit r/w) counter-mode: Set to False (the default) for Quadrature. Set to True for Step/Dir (in which
[23:16:29] <cradek> case Step is on the A pin and Dir is on the B pin).
[23:16:35] <skunkKandT> heh
[23:16:39] <skunkKandT> that might be why then...
[23:16:48] <cradek> yeah, that parameter is not what you think it is
[23:16:54] <skunkKandT> I just assumed.... funny
[23:17:08] <cradek> I was surprised that mesa could do non-x4
[23:17:50] <skunkKandT> well that explains that.
[23:18:14] <skunkKandT> cradek: how do you handle that?
[23:18:19] <skunkKandT> divide by 4?
[23:19:09] <cradek> yeah, just set the scale accordingly. it's kind of sucky because you get motion between clicks.
[23:19:26] <cradek> too bad every jog wheel in the world is non-x4
[23:19:32] <skunkKandT> rigth
[23:20:04] <cradek> non-x4 sounds easy until you start wondering about stuff like how index should work
[23:20:23] <cradek> I don't know if it can be reasonably faked in the driver, or if firmware support is needed to make it work right
[23:20:35] <skunkKandT> right
[23:23:39] <skunkKandT> it is funny that setting the counter mode made it count almost perfectly 100 counts per rev (vs 400)
[23:24:20] <skunkKandT> cradek: how do you have your encoder hooked in? are you scaling the increments? like for .001 you have .00025?
[23:24:30] <JT-Hardinge> cradek: are you sure you read encoder counter-mode?
[23:24:37] <JT-Hardinge> .counter-mode
[23:24:39] <JT-Hardinge> (Bit, RW) Set to False (the default) for Quadrature. Set to True for Up/Down or for single input on Phase A. Can be used for a frequency to velocity converter with a single input on Phase A when set to true.
[23:26:40] <cradek> setp scale.2.gain 0.00025
[23:26:40] <cradek> setp scale.3.gain -0.000125
[23:26:58] <cradek> yep I'm scaling it (for X, the increments are in diameter)
[23:27:23] <cradek> and clockwise moves in, instead of +x (out)
[23:27:43] <cradek> I'd go nuts if it worked opposite a manual lathe
[23:29:47] <cradek> hmm yep
[23:29:50] <cradek> on the mill too:
[23:29:50] <cradek> setp scale.0.gain .00025
[23:29:54] <cradek> net JogIncr touchy.jog.wheel.increment => scale.0.in
[23:29:57] <cradek> etc
[23:30:02] <cradek> PITA
[23:30:44] <cradek> but..... if you use ilowpass anyway, can't you just as easily scale it there?
[23:34:36] <skunkKandT> I would think
[23:34:51] <cradek> if you have high accel, ilowpass sure makes the wheel work nicer
[23:35:20] <cradek> otherwise - yes it really CAN move 0.010 and then stop again for each click of the wheel... bangity bangity bangity
[23:35:28] <cradek> (jr can, anyway)
[23:35:32] <skunkKandT> it is pretty smooth right now but we do have pretty low accels at the moment
[23:38:33] <skunkKandT> I have not looked yet - but are there enough pins in axis to use the gui for setting the jog wheel scales and stuff? (instead of having extra buttons and such - or a pyvcp pannel?
[23:40:01] <skunkKandT> (if that made sense)